DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant's arguments filed on 08/21/2025 have been fully considered but they are not persuasive.
Applicant argues on page 9 of the Applicant’s Remarks that “Applicant respectfully submits that each of Rozo et al., Zheng, and Asatani fails, alone or in combination, to teach or render obvious Applicant's claimed simple, reliable and robot-applied learning-model generation and AI-based weighing of a target fluid from a first container to a second container.”. The Examiner respectfully disagrees. Rozo teaches robot learning from demonstration with pouring liquid from a bottle into a glass is one such task. Rozo also teaches using one type of input for training which is the current joint value which is construed as the robot arm pose (See at least Page 230 Col 1 “V. RESULTS … The whole input space is composed of the selected subset of variables 0 and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t + 1.”). Rozo relies on Zheng in the rejection for the teachings of the second input which is the weight of the second container that is associated with the posture by the acquisition time (See at least Page 1 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are [Ɵ(t) Rotation angle at time t (degree) … ftarget Weight aimed to be poured in the receiving cup (lbf)”, discloses rotation angle which is construed as robot arm posture and weight of the target container which is construed as weight of the second container). Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Rozo with the teachings of Zheng and include the feature of the second input parameter which is the weight of the second container, thereby collecting more dataset for training which will provide precise results to control the robot arm.
The Applicant also argues on page of 9 of the Applicant’s Remarks that “Zheng is truly a scientific paper; but the disclosure/technique of Zheng is not practical in actual applications in the factory etc.”. The Examiner respectfully disagrees. The technique of Zheng is practically done as a project with dataset containing a total of 688 pouring sequences and their corresponding weight measurements (See at least Page 1 Col 2 “II. DATA AND PREPROCESSING”).
The Applicant further argues on page 9 of the Applicant’s Remarks that “Applicant thus respectfully submits that the combination of features claimed in Applicant's independent claims is unique in this respect, and is not easily conceivable from (i.e., is not obvious in view of) the applied references even for the person skilled in the art.”. The Examiner respectfully disagrees with the argument provided above. Therefore, the combination of Rozo and Zheng provided below teaches the Applicant’s claimed feature.
The same reasoning as applied to the independent claims above also apply to their corresponding dependent claims.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1, 3, 6, 8, 9, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Rozo et al. (L. Rozo, P. Jiménez and C. Torras, "Force-based robot learning of pouring skills using parametric hidden Markov models," 9th International Workshop on Robot Motion and Control, Kuslin, Poland, 2013, pp. 227-232) (Hereinafter Rozo) in view of Zheng (“Pouring Dynamics Estimation Using Gated Recurrent Units”).
Regarding Claim 1, Rozo teaches a method of generating a learning model for machine learning, the method performed by a hardware processor, the hardware processor reading previously set programs from a memory and executing the programs that have been read so as to perform (See at least Fig 1) steps of:
acquiring, for every acquisition time, learning data consisting of i) a posture of a robot arm provided with a plurality of arms connected to each other via a plurality of joints each of which rotatably links two of the arms so as to form the robot arm, each of the joints providing a rotation angle on an axis of each of the joints, the posture of the robot arm being defined as a set of the rotation angles each provided on each of all the axes of the robot arm, the robot arm holding a first container containing therein a target fluid and pouring the target fluid from the first container into a second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass, Page 230 Col 1 V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, Fig. 6. “Resulting 3-components PHMM trained with one demonstration of the task (four 100 ml drinks poured). Crosses show the skill parameter, for which the corresponding model is displayed using the same color. The PHMM satisfactorily encodes the task using a simple left-to-right topology. Robot joint values are given in degrees”), and …
generating, based on only the time-serial postures of the robot arm … acquired and stored in the further memory, the learning model, the learning model being given, as input thereto, only two types of information showing the posture of the robot arm (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”) … and outputting the information showing the posture of the robot arm at a second time which is a time later than the first time, the input to the learning model not including image information of the first container which is subjected to pouring the target fluid into the second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, discloses force and joint values as inputs to the learning model which is construed as not including image information of the first container as an input to the learning model).
However, Rozo does not explicitly spell out … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container;
storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other;
repeating the acquiring step and the storing step
… and the time serial weights of the second container … and the weight of the second container acquired at a first time, …
Zheng teaches … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”);
storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other (See at least age 1 Col 2 “II. DATA AND PREPROCESSING … Ɵ(t) Rotation angle at time t (degree) … ftarget Weight aimed to be poured in the receiving cup (lbf)”, discloses rotation angle which is construed as robot arm posture and weight of the target container which is construed as weight of the second container, also discloses data sets used for training, hence the data had to be collected in a memory);
repeating the acquiring step and the storing step (See at least “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements”, discloses 688 pouring sequences which means the sequences are repeated, hence examiner notes the acquiring and storing steps are to be repeated)…
… and the time serial weights of the second container … and the weight of the second container acquired at a first time (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”), …
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the robotic system of Rozo with the teachings of Zheng and include the feature of storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other and repeating the acquiring step and the storing step and also include the feature of weighing the second container when pouring from a first container and sent it as an input for machine modeling in order to receive output information showing the posture of the robot arm at a second time by processing two inputs, thereby collecting more dataset for training which will provide precise results to control the robot arm.
Regarding Claim 3, modified Rozo teaches all the elements of claim 1. Rozo further teaches the method according to claim 1, wherein the learning data includes, as data showing the posture of the robot arm, current values of motors provided in the respective joints of the robot arm, the current values serving as the load acting on the robot arm (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass, Page 230 Col 1 V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, Fig. 6. “Resulting 3-components PHMM trained with one demonstration of the task (four 100 ml drinks poured). Crosses show the skill parameter, for which the corresponding model is displayed using the same color. The PHMM satisfactorily encodes the task using a simple left-to-right topology. Robot joint values are given in degrees”, Page 230 Col 1 Section V. Results “In the proposed experimental task, namely the pouring skill, the robot perceptions are the wrench △={Fx Fy Fz Tx Ty Tz}, i.e., the sensed forces and torques in the robot's frame…”, Examiner notes Rozo implicitly discloses loads acting on robot arm involves current values of motors with respective joints of the robot arm).
Regarding Claim 6, Rozo teaches a method of weighing a target fluid, the method being performed by a hardware processor, the hardware processor reading previously set programs from a memory and executing the programs that have been read so as to perform (See at least Fig 1) steps of:
acquiring, for every acquisition time, learning data consisting of i) a posture of a robot arm provided with a plurality of arms connected to each other via a plurality of joints each of which rotatably links two of the arms so as to form the robot arm, each of the joints providing a rotation angle on an axis of each of the joints, the posture of the robot arm being defined as a set of the rotation angles each provided on each of all the axes of the robot arm, the robot arm holding a first container containing therein a target fluid and pouring the target fluid from the first container into a second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass, Page 230 Col 1 V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, Fig. 6. “Resulting 3-components PHMM trained with one demonstration of the task (four 100 ml drinks poured). Crosses show the skill parameter, for which the corresponding model is displayed using the same color. The PHMM satisfactorily encodes the task using a simple left-to-right topology. Robot joint values are given in degrees”), and …
generating, based on only the time-serial postures of the robot arm … acquired and stored in the further memory, the learning model, the learning model being given, as input thereto, only two types of information showing the posture of the robot arm (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”) … and outputting the information showing the posture of the robot arm at a second time which is a time later than the first time, the input to the learning model not including image information of the first container which is subjected to pouring the target fluid into the second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, discloses force and joint values as inputs to the learning model which is construed as not including image information of the first container as an input to the learning model); and
controlling movements of the robot arm based on the information showing the posture of the robot arm at the second time (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”).
However, Rozo does not explicitly spell out … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container;
storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other;
repeating the acquiring step and the storing step
… and the time serial weights of the second container … and the weight of the second container acquired at a first time, …
Zheng teaches … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”);
storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other (See at least Page 1 Col 2 “II. DATA AND PREPROCESSING … Ɵ(t) Rotation angle at time t (degree) … ftarget Weight aimed to be poured in the receiving cup (lbf)”, discloses rotation angle which is construed as robot arm posture and weight of the target container which is construed as weight of the second container, also discloses data sets used for training, hence the data had to be collected in a memory);
repeating the acquiring step and the storing step (See at least “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements”, discloses 688 pouring sequences which means the sequences are repeated, hence examiner notes the acquiring and storing steps are to be repeated)…
… and the time serial weights of the second container … and the weight of the second container acquired at a first time (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”), …
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the robotic system of Rozo with the teachings of Zheng and include the feature of storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other and repeating the acquiring step and the storing step and also include the feature of weighing the second container when pouring from a first container and sent it as an input for machine modeling in order to receive output information showing the posture of the robot arm at a second time by processing two inputs, thereby collecting more dataset for training and which will provide precise results to control the robot arm.
Regarding Claim 8, modified Rozo teaches all the elements of claim 6. Rozo further teaches the weighing method of claim 6, wherein the learning data includes, as data showing the posture of the robot arm, current values of motors provided in the respective joints of the robot arm, the current values serving as the load acting on the robot arm (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass, Page 230 Col 1 V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, Fig. 6. “Resulting 3-components PHMM trained with one demonstration of the task (four 100 ml drinks poured). Crosses show the skill parameter, for which the corresponding model is displayed using the same color. The PHMM satisfactorily encodes the task using a simple left-to-right topology. Robot joint values are given in degrees”, Page 230 Col 1 Section V. Results “In the proposed experimental task, namely the pouring skill, the robot perceptions are the wrench △={Fx Fy Fz Tx Ty Tz}, i.e., the sensed forces and torques in the robot's frame…”, Examiner notes Rozo implicitly discloses loads acting on robot arm involves current values of motors with respective joints of the robot arm).
Regarding Claim 9, Rozo teaches a system for weighing a target fluid, the system including a hardware processor and a memory, the hardware processor reading previously set programs from the memory and executing the programs (See at least Fig 3) that have been read to function as:
a state data acquisition unit configured to
acquire, for every acquisition time, learning data consisting of i) a posture of a robot arm provided with a plurality of arms connected to each other via a plurality of joints each of which rotatably links two of the arms so as to form the robot arm, each of the joints providing a rotation angle on an axis of each of the joints, the posture of the robot arm being defined as a set of the rotation angles each provided on each of all the axes of the robot arm, the robot arm holding a first container containing therein a target fluid and pouring the target fluid from the first container into a second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass, Page 230 Col 1 V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, Fig. 6. “Resulting 3-components PHMM trained with one demonstration of the task (four 100 ml drinks poured). Crosses show the skill parameter, for which the corresponding model is displayed using the same color. The PHMM satisfactorily encodes the task using a simple left-to-right topology. Robot joint values are given in degrees”), and …
generating, based on only the time-serial postures of the robot arm … acquired and stored in the further memory, the learning model, the learning model being given, as input thereto, only two types of information showing the posture of the robot arm (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”) … and outputting the information showing the posture of the robot arm at a second time which is a time later than the first time, the input to the learning model not including image information of the first container which is subjected to pouring the target fluid into the second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, discloses force and joint values as inputs to the learning model which is construed as not including image information of the first container as an input to the learning model);
a learning processing unit configured to input, to the learning model stored in the further memory, the information showing only the posture of the robot arm … and to make the learning model output information showing the posture of the robot arm at the second time (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, discloses force and joint values as inputs to the learning model which is construed as not including image information of the first container as an input to the learning model),
control unit configured to control movements of the robot arm based on the information showing the posture of the robot arm at the second time (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”).
However, Rozo does not explicitly spell out … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container;
store, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other;
repeat the acquiring step and the storing step
… and the time serial weights of the second container … and the weight of the second container acquired at a first time, …
and the weight of the second container acquired at the first time, …
Zheng teaches … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”);
store, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other (See at least age 1 Col 2 “II. DATA AND PREPROCESSING … Ɵ(t) Rotation angle at time t (degree) … ftarget Weight aimed to be poured in the receiving cup (lbf)”, discloses rotation angle which is construed as robot arm posture and weight of the target container which is construed as weight of the second container, also discloses data sets used for training, hence the data had to be collected in a memory);
repeat the acquiring step and the storing step (See at least “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements”, discloses 688 pouring sequences which means the sequences are repeated, hence examiner notes the acquiring and storing steps are to be repeated)…
… and the time serial weights of the second container … and the weight of the second container acquired at a first time (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”), …
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the robotic system of Rozo with the teachings of Zheng and include the feature of storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other and repeating the acquiring step and the storing step and also include the feature of weighing the second container when pouring from a first container and sent it as an input for machine modeling in order to receive output information showing the posture of the robot arm at a second time by processing two inputs, thereby collecting more dataset for training which will provide precise results to control the robot arm.
Regarding Claim 15, Rozo teaches a non-transitory computer-readable recording medium in which a computer program for generating a learning model is readably stored in advance (See at least Page 228 Col 1 Para 3 “Section V shows computational and robot execution results.”), the learning model being for machine learning directed to weighing a target fluid to be poured from a first container to a second container (See at least Title “Force-based Robot Learning of Pouring Skills”, Fig 3 ), the computer program enabling a computer to perform steps of:
acquiring, for every acquisition time, learning data consisting of i) a posture of a robot arm provided with a plurality of arms connected to each other via a plurality of joints each of which rotatably links two of the arms so as to form the robot arm, each of the joints providing a rotation angle on an axis of each of the joints, the posture of the robot arm being defined as a set of the rotation angles each provided on each of all the axes of the robot arm, the robot arm holding a first container containing therein a target fluid and pouring the target fluid from the first container into a second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass, Page 230 Col 1 V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, Fig. 6. “Resulting 3-components PHMM trained with one demonstration of the task (four 100 ml drinks poured). Crosses show the skill parameter, for which the corresponding model is displayed using the same color. The PHMM satisfactorily encodes the task using a simple left-to-right topology. Robot joint values are given in degrees”), and …
generating, based on only the time-serial postures of the robot arm … acquired and stored in the further memory, the learning model, the learning model being given, as input thereto, only two types of information showing the posture of the robot arm (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”) … and outputting the information showing the posture of the robot arm at a second time which is a time later than the first time, the input to the learning model not including image information of the first container which is subjected to pouring the target fluid into the second container (See at least Fig 3 shows robot learning to pour drinks from a bottle into a glass”, Page 230 Col 1 Para 3 - V. RESULTS “The whole input space is composed of the selected subset of variables δ and the current joint value q at time step t. Output is the desired robot joint position to be achieved at t+1. Thus, each training datapoint is defined as dmp = {Tx Fz Ty qt1 … qtN qt+11 … qt+1N}, where Nq is the number of joints of the robot. The fact of having included the robot state into the input vector is aimed at encapsulating the task dynamics, so that the next robot state depends not only on the force-based perceptions but also on the current robot joint values, which is in the line of a consistent representation of the skill”, discloses force and joint values as inputs to the learning model which is construed as not including image information of the first container as an input to the learning model).
However, Rozo does not explicitly spell out … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container;
storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other;
repeating the acquiring step and the storing step
… and the time serial weights of the second container … and the weight of the second container acquired at a first time, …
Zheng teaches … ii) a weight of the second container which changes time serially depending on the pouring of the target fluid from the first container (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”);
storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other (See at least age 1 Col 2 “II. DATA AND PREPROCESSING … Ɵ(t) Rotation angle at time t (degree) … ftarget Weight aimed to be poured in the receiving cup (lbf)”, discloses rotation angle which is construed as robot arm posture and weight of the target container which is construed as weight of the second container, also discloses data sets used for training, hence the data had to be collected in a memory);
repeating the acquiring step and the storing step (See at least “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements”, discloses 688 pouring sequences which means the sequences are repeated, hence examiner notes the acquiring and storing steps are to be repeated)…
… and the time serial weights of the second container … and the weight of the second container acquired at a first time (See at least Page 2 Col 2 “II. DATA AND PREPROCESSING The dataset contains a total of 688 pouring sequences and their corresponding weight measurements. Each motion sequence has seven feature dimensions, which for each timestamp of a motion sequence are … ftarget Weight aimed to be poured in the receiving cup (lbf)”), …
Therefore, it would have been obvious to one of the ordinary skill in the art before the effective filing date of the claimed invention to modify the robotic system of Rozo with the teachings of Zheng and include the feature of storing, in a further memory, the acquired posture of the robot arm and the acquired weight of the second container, together with the acquisition times at each of which the posture of the robot arm and the weight of the second container are made to correspond to each other and repeating the acquiring step and the storing step and also include the feature of weighing the second container when pouring from a first container and sent it as an input for machine modeling in order to receive output information showing the posture of the robot arm at a second time by processing two inputs, thereby collecting more dataset for training which will provide precise results to control the robot arm.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Briquet-Kerestedjian et al. (CN 111712356 A) teaches neural network trained to be based on velocity and torque data
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHAHEDA HOQUE whose telephone number is (571)270-5310. The examiner can normally be reached Monday-Friday 8:00 am- 5:00 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Ramon Mercado can be reached on 571-270-5744. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/SHAHEDA HOQUE/Examiner, Art Unit 3658
/TRUC M DO/Primary Examiner, Art Unit 3658