Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

{\alpha}-RACER: Real-Time Algorithm for Game-Theoretic Motion Planning and Control in Autonomous Racing using Near-Potential Function

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A learned near-potential function, maximized online, is shown to approximate a Nash equilibrium strategy for multi-car autonomous racing in real time.

desk verdict Genuine application of α-potential games to multi-car racing with a clean conditional theorem; the empirical evidence is self-referential and the online global-maximizer premise is unverified, but the paper deserves a serious referee. read the letter →

arxiv 2412.08855 v3 pith:OL45MSUC submitted 2024-12-12 cs.RO cs.GT

Dvij Kalaria , Chinmay Maheshwari , Shankar Sastry This is my paper · ORCID
classification cs.ROcs.GT MSC 91A1091A1593C85
keywords autonomousracingmulti-agentmotionplanningNashequilibriumalpha-potentialgamesmodelpredictivecontrolgame-theoreticreal-time
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that competitive multi-car racing can be solved in real time by replacing the online solution of a coupled dynamic game with the maximization of a single learned function, called a near-potential function. If the near-potential inequality holds, Proposition 3 guarantees that the maximizer is an approximate Nash equilibrium, so the ego car's chosen maneuver is a sound response to the other cars' strategies. The paper instantiates this idea for three-car racing with nonlinear tire dynamics, trains the function on 4,000 simulated races, and reports that the resulting policy wins more races than iterated best response and self-play reinforcement-learning opponents. A sympathetic reading is that this demonstrates a practical route from game-theoretic equilibrium concepts to real-time control in multi-agent racing.

What carries the argument

The load-bearing object is the dynamic $\alpha$-potential function, a scalar function satisfying, for every state $x$, agent $i$, and unilateral parameter change, the inequality $|(\Phi(x,\theta^i,\theta^{-i})-\Phi(x,\theta^{i'},\theta^{-i})) - (V^i(x,\theta^i,\theta^{-i})-V^i(x,\theta^{i'},\theta^{-i}))| \leq \alpha$. This inequality is what converts equilibrium computation into a single scalar maximization: Proposition 3 uses it to bound the Nash regret of any near-maximizer by $\lambda+\alpha$. The companion machinery is the five-parameter policy class $(q,\zeta,s_1,s_2,s_3)$, where $q$ sets tracking aggressiveness, $\zeta$ scales the race-line speed, and the $s_i$ shape overtaking and blocking offsets; this parametrization makes the strategy space finite-dimensional and semantically meaningful.

What would settle it

Run the same learned potential on a held-out batch of, say, 200 newly simulated races with random initial states and policy parameters, and evaluate the Definition 2 gap at those samples; if the gap exceeds the claimed 10 percent of the value-function range, the $\alpha$ in Proposition 3 is not the small value the paper reports. A second check is to compare the online maximizer's Nash regret not against the solver's own warm-started maximizer but against a dense grid search over all five policy parameters; if regret is not small, the algorithm's equilibrium claim rests on optimization error rather than potential accuracy.

Watch

Extended reading notes

Core claim

The paper's central claim is that the multi-car racing game admits a dynamic $\alpha$-potential function $\Phi(x,\theta;\phi^*)$ whose value differences approximate each agent's long-term value differences under unilateral policy changes, and that the online maximizer $\theta^* \in \arg\max_{\theta \in \Theta} \Phi(x_t,\theta;\phi^*)$ is therefore a $(\lambda+\alpha)$-approximate Nash equilibrium whenever the maximization is $\lambda$-near optimal. The game itself is an infinite-horizon discounted dynamic game in which each car's policy is a model-predictive controller tracking a reference trajectory parameterized by five interpretable parameters, and each car's one-step utility is its relative progress along the track. Working through this model, the paper shows numerically that the learned potential's approximation gap stays within roughly 10 percent of the value-function range on training samples and that the ego's Nash regret stays within roughly 3 percent, while the ego wins 61 to 91 of 99 races depending on the opponent algorithm.

Load-bearing premise

The guarantee depends on the learned potential network staying accurate at states and policy parameters it never saw during the 4,000 training races, but the paper checks the approximation gap only on training data, so a generalization failure would leave the online maximizer without its approximate-Nash certificate.

Editorial extensions

If this is right

  • Real-time equilibrium play reduces to a single scalar maximization per time step, so the online planner's computation is independent of the number of cars except through the potential network's input dimension.
  • Because Proposition 3 gives an explicit regret bound of $\lambda+\alpha$, any improvement in potential accuracy or optimization tightness directly tightens the approximate-Nash certificate.
  • The five-parameter policy class gives the equilibrium strategy an interpretable meaning: a chosen $\theta^*$ corresponds to concrete racing behaviors such as aggressive race-line tracking, speed perturbation, overtaking offset, or blocking response.
  • The numerical results suggest the approach beats short-horizon iterated best response and self-play RL in head-to-head three-car races, with win counts of 61, 52, 76, 73, and 91 out of 99 across the five opponent cases.
  • The framework is modular in the vehicle model, so the same potential-learning pipeline can be applied to different dynamics, including technologies like push-to-pass, without changing the equilibrium computation.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the near-potential property holds broadly, the offline-learn/online-maximize template could transfer to other real-time multi-agent settings with many players, such as intersection crossing or multi-drone racing, where solving coupled dynamic games online is infeasible; the paper does not claim this transfer.
  • A natural next test is a symmetric race in which all three cars run the same $\alpha$-RACER algorithm; the paper's 99-race evaluation pits one learned ego against fixed opponents, so it does not yet show what happens when every player is maximizing the same learned potential.
  • A direct way to separate optimization quality from potential quality is to compare the learned-potential maximizer against a dense grid search over the five policy parameters; the paper does not report this comparison.
  • One implicit consequence is that the value-function estimators $V^i$ are the foundation of the training target, so any bias in those estimators enters the learned potential's $\alpha$; accuracy of value estimation is the upstream lever for the whole pipeline.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes α-RACER, a two-phase algorithm for approximate Nash equilibrium computation in multi-car autonomous racing. In the offline phase, the authors simulate 4000 races, train value-function estimators V^i(·; υ) for each agent, and then fit a neural-network potential function Φ(·; φ*) by solving a sampled version of the semi-infinite program in Eq. (3), which enforces the α-potential inequality of Definition 2. In the online phase, the ego vehicle selects its policy parameter θ* by maximizing Φ(x_t, θ; φ*) via gradient ascent, and Proposition 3 is invoked to argue that this θ* is a (λ+α)-approximate Nash equilibrium. The paper reports a small approximation gap, small Nash regret, and superior winning rates against several baselines (variants with different discount factors or data sizes, iterated best response, and self-play RL).

Significance. If the central premises were verified, the approach would be a valuable step toward real-time game-theoretic motion planning for multi-agent racing with nonlinear vehicle dynamics. The modeling contributions—the policy parametrization for overtaking and blocking, the relative-progress utility, and the use of dynamic α-potential functions—are interesting and well aligned with the racing domain. Proposition 3 is clean and provides a useful bridge between potential maximization and approximate Nash equilibrium. However, the quantitative evidence in Section 4 does not yet establish the key premises: the learned value functions are used both as training targets and as evaluation yardsticks, the online optimizer lacks a global optimality certificate, and all reported gap/regret numbers are in-sample. The paper's significance is therefore conditional on a more rigorous validation of these assumptions.

major comments (3)
  1. [Section 3.2, Eq. (3) and Section 4.1, Figure 1(a)] The approximation gap in Figure 1(a) is computed using the same learned value functions V^i(·; υ) that served as training targets in Eq. (3), and only on the training samples. A potential function Φ can satisfy inequality (2) with respect to learned V^i while being far from an α-potential for the true value functions; hence the reported α does not measure closeness to true Nash equilibrium. To support the claim that Proposition 3 applies, the paper should evaluate the gap on held-out states and policy parameters, and ideally compare against value estimates obtained independently (for example, via Monte Carlo rollouts of the actual game), rather than the same fitted networks.
  2. [Section 4, Experimental Setup, and Proposition 3] The online phase uses gradient ascent with a fixed learning rate and warm-start from the previous time step to maximize Φ(x_t, θ; φ*) over a nonconcave neural network. Proposition 3 requires θ* to be within λ of the global maximizer of Φ, but no global optimality certificate, bound on the suboptimality λ, or multi-start/restart procedure is reported. Without such a bound, the online θ* need not be a (λ+α)-approximate Nash equilibrium for any meaningful λ. The paper should either provide a λ estimate (e.g., via a duality gap or randomized multi-start comparison) or explicitly frame the Nash regret result as conditional on the local optimization outcome.
  3. [Section 4.1, Figure 1(b)] The Nash regret plotted in Figure 1(b) is defined as max_{θ^i} V^i(x, θ^i, θ*^{-i}) - V^i(x, θ*^{i}, θ*^{-i}), and the text indicates that V^i is the learned estimator trained on the same 4000-race dataset. If so, this regret measures consistency between the potential maximizer and the learned value function, not distance to a true Nash equilibrium of the racing game. The paper should clarify exactly how V^i is obtained in this figure, how the max over θ^i is computed, and whether the reported regret is evaluated on held-out states or on the training distribution.
minor comments (6)
  1. [Definition 2, p. 7] The inequality in Definition 2 writes the bound as ≤ κ, but Proposition 3 uses α; this inconsistency should be fixed so the approximation parameter in the definition matches the one in the theorem.
  2. [Section 3.2 and Appendix D.1] The main text says the value network uses the same architecture as the potential function, but Appendix D.1 reports different sizes: (128,128,64) for the value network and (384,384,192) for the potential network. This discrepancy should be reconciled.
  3. [Section 2.2, near-collision dynamics] The hand-specified velocity reductions (v_x multiplied by 1/2 for the leading car and 1/3 for the trailing car) are ad hoc surrogates for collision outcomes. The paper should explicitly state this as a modeling assumption and discuss its sensitivity, since it can significantly affect the learned value functions and therefore the potential function.
  4. [Section 4.2, Table 1] For Cases I–III, the opponents are trained with the same method but with worse hyperparameters (lower/higher γ, fewer data). This does not isolate the algorithmic contribution of α-RACER; a table with the ego also evaluated under the worse hyperparameters would make the comparison more informative.
  5. [Section 4.2, first paragraph] The text says 'we conduct 99 races' and then describes cases with 99 races each; it would be clearer to state explicitly that each case consists of 99 races (or, if not, to correct the sentence).
  6. [Section 2.2, MPC formulation] Equation (1f) uses w_max as the track width, but the surrounding text says 'w_max is the track length'; the notation should be made consistent (track width is meant from the constraint |p_y| ≤ w_max/2).

Circularity Check

2 steps flagged · score 6.0 of 10

The reported α-gap and Nash regret are in-sample consistency checks: the potential is fitted to learned value functions V^i in Eq. (3), and then validated with those same learned V^i on the training samples, so the near-Nash claims partially reduce to the fitting objective.

  1. fitted input called prediction [Section 3.2, Eq. (3); Section 4.1, Figure 1(a)]
    "we cast the problem of learning potential function as a semi-infinite program ... min_{y,ϕ} y s.t. |(Φ(x, θi, θ−i; ϕ) − Φ(x, θi′, θ−i; ϕ)) − (V i(x, θi, θ−i; υ) − V i(x, θi′, θ−i; υ))| ≤ y ... We observe that the approximation gap across all states and policy parameters used in the training samples remains within 10% of the value function’s range."

    Equation (3) defines the training objective as minimizing exactly the worst-case discrepancy between differences of Φ and differences of the learned value estimates V^i. Figure 1(a) then reports that same discrepancy on the training samples. This is a report of the fitted training loss, not an independent measurement of α for the true game. It cannot, by itself, support the paper's conclusion that 'the dynamic game admits an α-potential function with small α', because the 'value function' whose range is used is the learned estimator, and the states and parameters are the ones used to fit Φ.

  2. fitted input called prediction [Section 4.1, Figure 1(b)]
    "Figure 1(b) shows the Nash regret for the ego agent, defined as maxθi V i(x, θi, θ∗,−i) − V i(x, θ∗,i, θ∗,−i), where θ∗ is the optimizer of the potential function with the starting state x. ... This dataset is used to first train value function estimators V 1, V 2, and V 3 for each of the cars. These are then used to learn Φ using (3)."

    The V^i appearing in the Nash regret is the same learned value-function estimate used to define the training target in Eq. (3). Therefore the small regret measures consistency between the fitted potential maximizer and the fitted value network on the same simulated dataset, not the distance to a true Nash equilibrium of the racing game. If V^i is inaccurate, both the fitted potential and the reported regret are inaccurate in the same direction, so the reported 3% regret is not independent evidence for the paper's near-Nash-equilibrium claim.

full rationale

The proof of Proposition 3 in Appendix A is mathematically self-contained: given a genuine α-potential function and a λ-global maximizer, the (λ+α)-approximate Nash guarantee follows directly from Definition 2. The citations to overlapping prior work on α-potential games (Guo et al. 2023; Maheshwari et al. 2024; Guo et al. 2024) are not load-bearing for this proposition because the proof is supplied in the paper itself. The circularity is in the empirical validation chain: the potential function is fitted to learned value estimates V^i, and the headline numbers in Figure 1 are computed with those same learned estimates on the same training samples. Thus the 'approximation gap' is the training objective, and the 'Nash regret' is an internal-consistency measure. The race-win comparisons against IBR and self-play RL are external benchmarks, so the method's competitive performance claim has independent content, but that does not rescue the specific claim that the learned maximizer is validated as an approximate Nash equilibrium. Separately, the online algorithm uses gradient ascent with warm-start on a nonconcave network, so the λ premise of Proposition 3 is not certified; this is a correctness risk, not a circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim is carried by two neural networks fitted to one simulated dataset and by several ad hoc modeling choices. The theoretical Proposition 3 is conditional on an accurate α-potential function, and the ledger shows that the paper obtains that function by fitting rather than by derivation.

free parameters (4)
  • Potential network weights ϕ = not reported
    Trained offline on 4000 simulated races via Eq (3) to make Φ approximate utility differences; the online maximization and all reported guarantees depend on these fitted weights.
  • Value network weights υ = not reported
    Trained on the same simulated races to estimate V_i; used both as training target in Eq (3) and as the evaluation oracle for Nash regret in Figure 1(b), so errors in υ propagate into reported guarantees.
  • Discount factor γ = 0.99 for ego, 0.98 and 0.995 for opponents
    Chosen hyperparameter that changes the effective horizon and the fitted value and potential functions; sensitivity is only tested through opponent variations, not for the ego.
  • Collision and track-boundary penalty factors = 1/2, 1/3, 1/2
    Ad hoc dynamics modifications in Section 2.2: nearby cars lose velocity by factors 1/2 and 1/3, out-of-track cars by 1/2. These shape the simulated data used for both training and evaluation.
assumptions (4)
  • ad hoc to paper An α-potential function exists for the racing game and is representable by the chosen neural network class.
    No theorem establishes existence for this game with this policy parametrization; the paper learns it from data in Section 3.2.
  • domain assumption The finite training dataset covers enough of the state and policy space for the learned potential and value functions to generalize to online states.
    The sampled semi-infinite program in Eq (3) is solved on randomly selected states and parameters; no coverage or generalization guarantee is given in Sections 3.2 and 4.1.
  • domain assumption The restricted policy class Θ = {θ = (q, ζ, s1, s2, s3)} contains strategies close to the true equilibrium of the unrestricted racing game.
    Proposition 3 can only certify approximate Nash equilibrium within Θ; the paper does not bound the restriction error to the full policy space in Section 2.2.
  • ad hoc to paper The near-collision velocity reduction heuristics are a faithful surrogate for real race outcomes.
    Section 2.2 replaces collision with deterministic speed penalties; all training and evaluation data inherit this model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of {\alpha}-RACER: Real-Time Algorithm for Game-Theoretic Motion Planning and Control in Autonomous Racing using Near-Potential Function." pith.science (2026). https://pith.science/paper/OL45MSUC

@misc{pith2026241208855,
  author       = {Pith},
  title        = {Pith review of: \alpha-RACER: Real-Time Algorithm for Game-Theoretic Motion Planning and Control in Autonomous Racing using Near-Potential Function},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OL45MSUC}},
  note         = {Machine review of arXiv:2412.08855}
}
read the original abstract

Autonomous racing extends beyond the challenge of controlling a racecar at its physical limits. Professional racers employ strategic maneuvers to outwit other competing opponents to secure victory. While modern control algorithms can achieve human-level performance by computing offline racing lines for single-car scenarios, research on real-time algorithms for multi-car autonomous racing is limited. To bridge this gap, we develop game-theoretic modeling framework that incorporates the competitive aspect of autonomous racing like overtaking and blocking through a novel policy parametrization, while operating the car at its limit. Furthermore, we propose an algorithmic approach to compute the (approximate) Nash equilibrium strategy, which represents the optimal approach in the presence of competing agents. Specifically, we introduce an algorithm inspired by recently introduced framework of dynamic near-potential function, enabling real-time computation of the Nash equilibrium. Our approach comprises two phases: offline and online. During the offline phase, we use simulated racing data to learn a near-potential function that approximates utility changes for agents. This function facilitates the online computation of approximate Nash equilibria by maximizing its value. We evaluate our method in a head-to-head 3-car racing scenario, demonstrating superior performance compared to several existing baselines.

Figures

Figures reproduced from arXiv: 2412.08855 by the authors.

Figure 1
Figure 1. Histogram of (a) Relative approximation gap of potential function (b) Nash regret [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Potential values and the trajectories at a given joint state for different (a) [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Example overtake in a race MPC vs IBR. Opponent (IBR) overtakes from [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Track and the starting position regions Appendix C. Dynamic Bicycle Model For any car i, we denote its mass by mi , its moment of inertia in the vertical direction about the center of mass by I i z , the distance between the center of mass (COM) and its front wheel by …

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Expert Knowledge-driven Reinforcement Learning for Autonomous Racing via Trajectory Guidance and Dynamics Constraints

    cs.RO 2026-03 conditional novelty 5.0 of 10

    Embedding a minimum-curvature racing line and CBF dynamics constraints into curriculum RL yields faster, more stable simulated laps than PPO, DDPG, and trajectory-aided baselines on Tempelhof.

Reference graph

Works this paper leans on

24 extracted references · 17 canonical work pages · cited by 1 Pith paper

  1. [1]

    Deepracer: Autonomous racing platform for experimentation with sim2real reinforcement learning

    Bharathan Balaji, Sunil Mallya, Sahika Genc, Saurabh Gupta, Leo Dirac, Vineet Khare, Gourav Roy, Tao Sun, Yunzhe Tao, Brian Townsend, et al. Deepracer: Autonomous racing platform for experimentation with sim2real reinforcement learning. In 2020 IEEE international conference on robotics and automation (ICRA), pages 2746–2754. IEEE,

    work page 2020

  2. [6]

    Markov α-potential games: Equilibrium approximation and regret analysis

    Xin Guo, Xinyu Li, Chinmay Maheshwari, Shankar Sastry, and Manxi Wu. Markov α-potential games: Equilibrium approximation and regret analysis. arXiv preprint arXiv:2305.12553 , 3,

    arXiv

  3. [8]

    2019.1631455

    doi: 10.1080/00423114. 2019.1631455. James Herman, Jonathan Francis, Siddha Ganju, Bingqing Chen, Anirudh Koul, Abhinav Gupta, Alexey Skabelkin, Ivan Zhukov, Max Kumskoy, and Eric Nyberg. Learn-to-race: A multimodal control environment for autonomous racing. In proceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 9793–9802,

    work page arXiv 2019

  4. [10]

    Rapid: Autonomous multi-agent racing using con- strained potential dynamic games

    Yixuan Jia, Maulik Bhatt, and Negar Mehr. Rapid: Autonomous multi-agent racing using con- strained potential dynamic games. In 2023 European Control Conference (ECC) , pages 1–8. IEEE,

    work page 2023

  5. [11]

    Game-theoretic model predictive control with data-driven identification of vehicle model for head-to-head autonomous racing

    Chanyoung Jung, Seungwook Lee, Hyunki Seong, Andrea Finazzi, and David Hyunchul Shim. Game-theoretic model predictive control with data-driven identification of vehicle model for head-to-head autonomous racing. arXiv preprint arXiv:2106.04094,

    arXiv

  6. [12]

    Dvij Kalaria, Qin Lin, and John M. Dolan. Adaptive planning and control with time-varying tire models for autonomous racing using extreme learning machine. ArXiv, abs/2303.08235, 2023a. URL https://api.semanticscholar.org/CorpusID:257532643. Dvij Kalaria, Qin Lin, and John M Dolan. Towards optimal head-to-head autonomous racing with curriculum reinforceme...

    work page Pith review arXiv

  7. [13]

    Potential ilqr: A potential-minimizing controller for planning multi-agent interactive trajectories

    Talha Kavuncu, Ayberk Yaraneri, and Negar Mehr. Potential ilqr: A potential-minimizing controller for planning multi-agent interactive trajectories. arXiv preprint arXiv:2107.04926,

    arXiv

  8. [16]

    Autonomous racing using learning model predictive control

    Ugo Rosolia, Ashwin Carvalho, and Francesco Borrelli. Autonomous racing using learning model predictive control. In 2017 American control conference (ACC), pages 5115–5120. IEEE,

    work page 2017

Show all 24 references
  1. [17]

    Autonomous drone racing with deep reinforcement learning

    Yunlong Song, Mats Steinweg, Elia Kaufmann, and Davide Scaramuzza. Autonomous drone racing with deep reinforcement learning. In 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 1205–1212. IEEE,

    2021

  2. [18]

    Hierarchical control for head-to-head autonomous racing

    Rishabh Saumil Thakkar, Aryaman Singh Samyal, David Fridovich-Keil, Zhe Xu, and Ufuk Topcu. Hierarchical control for head-to-head autonomous racing. arXiv preprint arXiv:2202.12861 ,

    arXiv

  3. [20]

    Game theoretic motion planning for multi-robot racing

    13 KALARIA MAHESHWARI ∗ SASTRY Zijian Wang, Riccardo Spica, and Mac Schwager. Game theoretic motion planning for multi-robot racing. In Distributed Autonomous Robotic Systems: The 14th International Symposium , pages 225–238. Springer, 2019b. Zijian Wang, Tim Taubner, and Mac ...

    arXiv

  4. [21]

    Game-theoretic objective space planning

    Hongrui Zheng, Zhijun Zhuang, Johannes Betz, and Rahul Mangharam. Game-theoretic objective space planning. arXiv preprint arXiv:2209.07758,

    arXiv

  5. [22]

    A sequential quadratic programming approach to the solution of open-loop generalized nash equilibria

    Edward L Zhu and Francesco Borrelli. A sequential quadratic programming approach to the solution of open-loop generalized nash equilibria. In 2023 IEEE International Conference on Robotics and Automation (ICRA), pages 3211–3217. IEEE,

    2023

  6. [23]

    A sequential quadratic programming approach to the solution of open-loop generalized nash equilibria for autonomous racing

    Edward L Zhu and Francesco Borrelli. A sequential quadratic programming approach to the solution of open-loop generalized nash equilibria for autonomous racing. arXiv preprint arXiv:2404.00186,

    arXiv

  7. [24]

    Proof of Proposition 3 Consider a policy parameter θ∗ such that Φ(x, θ∗) ≥ maxθ∈Θ Φ(x, θ) − λ

    14 GAME -T HEORETIC MOTION PLANNING AND CONTROL IN AUTONOMOUS RACING Appendix A. Proof of Proposition 3 Consider a policy parameter θ∗ such that Φ(x, θ∗) ≥ maxθ∈Θ Φ(x, θ) − λ. For any θi ∈ Θi, V i(x, θ∗,i, θ∗,−i) − V i(x, θi, θ∗,−i) ≥ Φ(x, θ∗,i, θ∗,−i) − Φ(x, θi, θ∗,−i) − α ≥ ...

    2020

  8. [2015]

    Game theoretic planning for self-driving cars in competitive scenarios

    Mingyu Wang, Zijian Wang, John Talbot, J Christian Gerdes, and Mac Schwager. Game theoretic planning for self-driving cars in competitive scenarios. In Robotics: Science and Systems, pages 1–9, 2019a. Mingyu Wang, Negar Mehr, Adrien Gaidon, and Mac Schwager. Game-theoretic pla...

    2020

  9. [2016]

    11 KALARIA MAHESHWARI ∗ SASTRY Achin Jain and Manfred Morari

    URL https://api.semanticscholar.org/CorpusID: 16153365. 11 KALARIA MAHESHWARI ∗ SASTRY Achin Jain and Manfred Morari. Computing the racing line using Bayesian optimization.2020 59th IEEE Conference on Decision and Control (CDC), pages 6192–6197,

    2020

  10. [2017]

    Algames: A fast solver for constrained dynamic games

    Simon Le Cleac’h, Mac Schwager, and Zachary Manchester. Algames: A fast solver for constrained dynamic games. arXiv preprint arXiv:1910.09713,

    1910 arXiv

  11. [2019]

    Efficient iter- ative linear-quadratic approximations for nonlinear multi-player general-sum differential games

    David Fridovich-Keil, Ellis Ratner, Lasse Peters, Anca D Dragan, and Claire J Tomlin. Efficient iter- ative linear-quadratic approximations for nonlinear multi-player general-sum differential games. In 2020 IEEE international conference on robotics and automation (ICRA) , page...

    2020

  12. [2020]

    Alexander Buyval, Aidar Gabdulin, Ruslan Mustafin, and Ilya Shimchik

    URL https://api.semanticscholar.org/CorpusID: 218487458. Alexander Buyval, Aidar Gabdulin, Ruslan Mustafin, and Ilya Shimchik. Deriving overtaking strat- egy from nonlinear model predictive control for a race car. In 2017 IEEE/RSJ international conference on intelligent robots...

    2017

  13. [2021]

    Multi-hypothesis inter- actions in game-theoretic motion planning

    Forrest Laine, David Fridovich-Keil, Chih-Yuan Chiu, and Claire Tomlin. Multi-hypothesis inter- actions in game-theoretic motion planning. In 2021 IEEE International Conference on Robotics and Automation (ICRA), pages 8016–8023. IEEE,

    2021

  14. [2022]

    Learning model predictive control for competitive autonomous racing

    Lukas Brunke. Learning model predictive control for competitive autonomous racing. ArXiv, abs/2005.00826,

    2005 arXiv

  15. [2023]

    Anα-potential game framework forn-player dynamic games

    Xin Guo, Xinyu Li, and Yufei Zhang. Anα-potential game framework forn-player dynamic games. arXiv preprint arXiv:2403.16962,

    arXiv

  16. [2024]

    Enhancing game- theoretic autonomous car racing using control barrier functions

    Gennaro Notomista, Mingyu Wang, Mac Schwager, and Magnus Egerstedt. Enhancing game- theoretic autonomous car racing using control barrier functions. In 2020 IEEE international conference on robotics and automation (ICRA), pages 5393–5399. IEEE,

    2020

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.