DETAILED CORRESPONDENCE
This action is in response to the filing of the Arguments and Amendments on 05/26/2026.
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 .
Claim Objections
The claims below are objected to because of the listed informalities, see below. Appropriate correction is required.
Claim 2, lines 9-10 should read ‘calculating the first reward corresponding to the first driving path information’.
Claim 5, lines 9-10 should read ‘calculating the second reward corresponding to the second driving path information’.
Claim 10, lines 8-9 should read ‘calculate the first reward corresponding to the first driving path information’.
Claim 13, lines 8-9 should read ‘calculate the second reward corresponding to the second driving path information’.
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.
Claim(s) 1 and 9 are rejected under 35 U.S.C. 103 as being unpatentable over Lee (US 2021/0380099) in view of Kazemi (US 20180292824).
Claim 1, Lee discloses a method for controlling an autonomous vehicle, the method comprising:
in response to a determination that a plurality of first driving path information corresponding to first state information is acquired, calculating a first reward corresponding to at least a portion of the first driving path information based on a result of performing a driving simulation according to the first driving path information [see Lee – p0003 – p0005, p0021, p0043 – p0044, path planning and control to account for position uncertainty for autonomous machine applications. Systems and methods are disclosed that generate and select candidate paths for a vehicle using an uncertainty representation for the vehicle; An uncertainty representation can be determined in real-time so the position of the trailer of the autonomous vehicle can be used in path planning and selection determinations. For example, a path generator (as part of a candidate path manager) may generate any number of paths having associated target locations (e.g., determined using any number of path generation techniques), and the uncertainty representations for the autonomous vehicle may be generated for the target locations to represent potential future footprints of the vehicle at the locations. As an example, two or more path candidates may be generated at each time step—e.g., a first path candidate may correspond to the autonomous vehicle staying in the current lane and a second path candidate may correspond to a lane change. These two paths may have been obtained by solving a constraint optimization problem such that each of the paths satisfy the constraints enforced thereon, such as a stochastic distance constraint. A path selector—e.g., of a planning layer—may thus select one of the paths to follow, and this information may be passed to control components of the vehicle for controlling the vehicle according to the path. The method 700, at block B706, includes computing costs for each of the plurality of candidate paths based on enforcing a constraint(s) in view of the uncertainty representations; the safety or collision avoidance consideration, may factor in to the final determination of a path for the vehicle 800. This consideration may be used to filter out paths, penalize (e.g., apply or attribute a negative or lower weight value to) paths where collision or possible collision events are predicted between the vehicle 800 and one or more actors, reward (e.g., apply or attribute a positive or higher weight value to) paths where an absence of a collision or possible collision event is predicted];
in response to a determination that second state information is acquired, generating at least one second driving path information corresponding to the second state information through the driving path generation network, calculating a second reward corresponding to at least a portion of the second driving path information based on a result of performing a driving simulation according to the second driving path information [see Lee – p0003 – p0005, p0021, p0043 – p0044, p0074, Figs, 3, 4A, 4B, 5 and 7 - path planning and control to account for position uncertainty for autonomous machine applications. Systems and methods are disclosed that generate and select candidate paths for a vehicle using an uncertainty representation for the vehicle; An uncertainty representation can be determined in real-time so the position of the trailer of the autonomous vehicle can be used in path planning and selection determinations. For example, a path generator (as part of a candidate path manager) may generate any number of paths having associated target locations (e.g., determined using any number of path generation techniques), and the uncertainty representations for the autonomous vehicle may be generated for the target locations to represent potential future footprints of the vehicle at the locations several paths (a first and a second is taught). As an example, two or more path candidates may be generated at each time step—e.g., a first path candidate may correspond to the autonomous vehicle staying in the current lane and a second path candidate may correspond to a lane change. These two paths may have been obtained by solving a constraint optimization problem such that each of the paths satisfy the constraints enforced thereon, such as a stochastic distance constraint. A path selector—e.g., of a planning layer—may thus select one of the paths to follow, and this information may be passed to control components of the vehicle for controlling the vehicle according to the path. The method 700, at block B706, includes computing costs for each of the plurality of candidate paths based on enforcing a constraint(s) in view of the uncertainty representations; the safety or collision avoidance consideration, may factor in to the final determination of a path for the vehicle 800. This consideration may be used to filter out paths, penalize (e.g., apply or attribute a negative or lower weight value to) paths where collision or possible collision events are predicted between the vehicle 800 and one or more actors, reward (e.g., apply or attribute a positive or higher weight value to) paths where an absence of a collision or possible collision event is predicted];
and in response to a determination that test state information is acquired, generating test driving path information corresponding to the test state information through the trained driving path generation network, and controlling the autonomous vehicle using the test driving path information [see p0074, at block B708, includes selecting a candidate path from the plurality of candidate paths based on the costs. For example, comfort, safety procedure execution analysis, obeying rules of the road, and/or other considerations may be factored in to determine which of the selected paths from the system 100 is a best or most suitable path for the vehicle 800 at a current time step].
Lee does not specifically teach and performing initial training of a driving path generation network based on at least a portion of the first reward; and performing fine tuning on the driving path generation network by training the driving path generation network based on at least a portion of the second reward.
However, Kazemi discloses automatic tuning of a plurality of gains of one or more cost functions used by a motion planning system of an autonomous vehicle.
Kazemi teaching, the automatic tuning system is configured to receive an autonomous motion plan generated by the autonomous vehicle motion planning system based at least in part on the data collected during the previous humanly-controlled vehicle driving session. The optimization planner optimized the one or more cost functions to generate the autonomous motion plan. The automatic tuning system is configured to obtain a humanly-executed motion plan that was executed during the previous humanly-controlled vehicle driving session. The automatic tuning system is configured to optimize an objective function to determine an adjustment to at least one of the plurality of gains. The objective function provides an objective value based at least in part on a difference between a first total cost obtained by input of the humanly-executed motion plan into the one or more cost functions of the autonomous vehicle motion planning system and a second total cost obtained by input of the autonomous motion plan into the one or more cost functions of the autonomous vehicle motion planning system [see Summary of Inv].
In Kazemi, the first initial training driving path is the humanly-controlled vehicle driving session, which provides a first total cost. Kazemi teaches the cost function(s) are quadratic, linear, or a combination thereof. Furthermore, in some implementations, the cost function(s) can include a portion that provides a reward rather than a cost. For example, the reward can be of opposite sign to cost(s) provided by other portion(s) of the cost function. An example, rewards can be provided for distance traveled, velocity, or other forms of progressing toward completion of a route [see p0043 – p0044].
Kazemi teaching the reward can be the opposite of cost. For example a cost function can be a motion planning that is over-time, not a reward [see p0101 – p0113].
Figures 2 - 5 teach the automatic tuning system (see 420) is configured to obtain a humanly-executed motion plan that provides a first total cost that was executed during the previous humanly-controlled vehicle driving session and a second total cost associated with the autonomous motion plan 506.
According to Kazemi, the total cost can be based at least in part on one or more cost functions 304. In one example implementation, the total cost equals the sum of all costs minus the sum of all rewards and the optimization planner attempts to minimize the total cost [see p0114 – p0136].
It would have been obvious before the effective date of the claimed invention to one of ordinary skill in the art to modify the device in Lee, to include and performing initial training of a driving path generation network based on at least a portion of the first reward; and performing fine tuning on the driving path generation network by training the driving path generation network based on at least a portion of the second reward, as suggested and taught by Kazemi, with a reasonable expectation of success, for the purpose of providing networks designed to select the "best path per reward" utilize reinforcement learning (RL) and graph-based algorithms to maximize utility, such as minimizing travel time, energy consumption, or risk, while accumulating rewards (e.g., speed bonuses, safety points). These systems, particularly in autonomous driving (AV), operate by creating a reward function that guides vehicle behavior—like lane keeping or overtaking—to navigate complex environments, and reduce costs such as over-time, approaching lane boundaries, etc.
Claim 9 is similarly rejected as Claim 1, see above.
Claim(s) 2 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Lee (US 2021/0380099) in view of Kazemi (US 20180292824), and Trajectory Planning in Frenet Frame via Multi-Objective Optimization, IEEE, 2023 (hereinafter referred to as Frenet).
Claim 2, Lee discloses the method of claim 1, but is silent to wherein calculating the first reward includes: generating the first driving path information corresponding to the first state information based on road information on a Frenet frame;
generating first mapping driving path information by mapping the first driving path information to a Cartesian frame;
and calculating the first reward corresponding to first driving path information on the Frenet frame that has been mapped to the first mapping driving path information on the Cartesian frame based on a result of performing a driving simulation according to the first mapping driving path information.
However, Frenet teaches wherein calculating the first reward includes: generating the first driving path information corresponding to the first state information based on road information on a Frenet frame [see Frenet, path planning technology is broadly classified into two categories: the first category is global path planning, which aims to find the optimal or suboptimal path from the starting point to the destination point. The second category is local path planning, which involves obtaining environmental information through sensors in unknown or partially unknown environments, allowing autonomous driving vehicles to obtain a collision-free executable optimal planned path [see page 2, Col. 1].
Further disclosing, optimal trajectory is selected by minimizing a predefined cost function formulated for optimal path planning, taking into account comfort, safety, and road center line deviation. See Figure 4, The framework of the proposed algorithm consists of two main stages: trajectory generation in the Frenet frame and optimal trajectory selection. Figure 4 shows that collisions can occur on the generated trajectories, and that there are limitations on the vehicle’s motion and dynamic characteristics. To enhance the system’s response time, trajectories that fail to meet the constraints are eliminated through trajectory checking. The remaining trajectories are then presented as candidate paths for the subsequent module to choose the best path. After the trajectory check, a set of candidate trajectories is generated. However, the number of candidates remains large, and we must choose a single trajectory to follow. To do so, we develop a cost function that assesses each candidate [see Fig 4, page 7, Section E. Cost Function].
Frenet, discloses as shown in Figure 3, the ego vehicle often needs to adjust its driving trajectory due to the presence of other vehicles and obstacles, instead of strictly following the reference line (i.e., the road center line). When in the Cartesian frame, it can describe the current state of the ego vehicle; FIGURE 3. Transformation from Frenet frame to cartesian frame, the total loss of each trajectory is calculated, and the trajectory with the minimum total loss is chosen as the optimal trajectory. The cost function is composed of three indicators: comfort, trajectory safety, and trajectory anti-deviation [see Figs, 3 – 5 and pages 5 – 8]. In the simulation, a straight road, a curvy road, an intersection scenario and a ‘‘U’’ shaped road are built in a Python environment, and several static obstacles of different sizes are set up on the roads. The experiments in this paper are divided into two parts: the first part analyzes the impact of different cost functions on trajectory generation [see Section IV, page 8].
Frenet also teaches the total loss of each trajectory is calculated, and the trajectory with the minimum total loss is chosen as the optimal trajectory. The Examiner interprets this calculation of min. loss to be a reward, as Frenet teaches reinforcement learning methods employing diverse reward strategies [See page 2, Section A].
It would have been obvious before the effective date of the claimed invention to one of ordinary skill in the art to modify the device in Lee, to include wherein calculating the first reward includes: generating the first driving path information corresponding to the first state information based on road information on a Frenet frame; generating first mapping driving path information by mapping the first driving path information to a Cartesian frame; and calculating the first reward corresponding to first driving path information on the Frenet frame that has been mapped to the first mapping driving path information on the Cartesian frame based on a result of performing a driving simulation according to the first mapping driving path information, as suggested and taught by Frenet, with a reasonable expectation of success, for the purpose of providing candidate paths free of collisions, also proposes a method to assess the safety of candidate trajectories based on their distance from obstacles, and the evaluation of the safety values of the candidate paths is In the candidate trajectory selection stage; With a new cost function to select the optimal trajectory, this cost function is designed to comprehensively consider comfort and safety.
Claim 10 is similarly rejected as Claim 2, see above.
Allowable Subject Matter
Claim 3 – 8, 11 – 16 objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
The examiner has pointed out particular references contained in the prior art of record in the body of this action for the convenience of the applicant. Although the specified citations are representative of the teachings in the art and are applied to the specific limitations within the individual claim, other passages and figures may apply as well. Applicant should consider the entire prior art as applicable as to the limitations of the claims. It is respectfully requested from the applicant, in preparing the response, to consider fully the entire references as potentially teaching all or part of the claimed invention, as well as the context of the passage as taught by the prior art or disclosed by the examiner.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to RENEE LAROSE whose telephone number is (313)446-4856. The examiner can normally be reached on Monday - Friday 8:30am - 5:00pm EST.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Abby Lin can be reached on (571) 270-3976. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/Renee LaRose/Examiner, Art Unit 3657
/ABBY LIN/ Supervisory Patent Examiner, Art Unit 3657
Prosecution Insights