aiqbench analysis
The paper uses a graph neural network trained by imitation learning to pick avoidance modes for multimodal model predictive control, cutting the exponential cost of enumerating modes to at most two optimal control solves per step. The reported results are a 93% success rate relative to the expert and an 18.6 ms single-thread runtime versus 35.5 ms for the expert on sixteen threads, based on 100 simulated mixed-traffic runs. Readers should note that the safety guarantee comes from the control barrier function-constrained optimal control problem rather than the learned component, which only proposes modes.
Imitation Learning of Avoidance Modes for Efficient Multimodal Model Predictive Control
arXiv:2610.16326v1 Announce Type: new Abstract: Multimodal model predictive control (MM-MPC) avoids local minima among moving obstacles by solving one optimal control problem (OCP) per combination of avoidance sides, so its computational cost grows exponentially with the number of decision-critical obstacles. We propose SideNet-MPC, which replaces this enumeration with a mode decision learned by imitating MM-MPC as an expert. A graph neural network is trained to predict the expert's minimum-cost mode, one avoidance side per obstacle. Its edges encode, for each obstacle pair, the gap predicted for the moment the vehicle first reaches either obstacle, so a neighbor that blocks a preferred passage can reverse the proposed side. The expert's selection rule compares its minimum-cost mode with the previously executed mode; SideNet-MPC keeps this rule and substitutes the network proposal for the minimum-cost mode. Hence at most two OCPs are solved per control period, and every applied input still comes from an OCP with control barrier function constraints rather than from the network. In $100$ seed-matched closed-loop Monte Carlo runs of mixed traffic, SideNet-MPC reaches $93\%$ of the expert's success rate and needs \SI{18.6}{ms} per control period on a single thread, 48\% less than the expert's \SI{35.5}{ms} on sixteen parallel threads. In a controlled gap sweep, its proposed side reverses in the same gap interval as the expert's, whereas proposers without pairwise edges, trained on the same data, never reverse and all deadlock at the blocked gates of a deterministic scene.
Abstract:Multimodal model predictive control (MM-MPC) avoids local minima among moving obstacles by solving one optimal control problem (OCP) per combination of avoidance sides, so its computational cost grows exponentially with the number of decision-critical obstacles. We propose SideNet-MPC, which replaces this enumeration with a mode decision learned by imitating MM-MPC as an expert. A graph neural network is trained to predict the expert's minimum-cost mode, one avoidance side per obstacle. Its edges encode, for each obstacle pair, the gap predicted for the moment the vehicle first reaches either obstacle, so a neighbor that blocks a preferred passage can reverse the proposed side. The expert's selection rule compares its minimum-cost mode with the previously executed mode; SideNet-MPC keeps this rule and substitutes the network proposal for the minimum-cost mode. Hence at most two OCPs are solved per control period, and every applied input still comes from an OCP with control barrier function constraints rather than from the network. In $100$ seed-matched closed-loop Monte Carlo runs of mixed traffic, SideNet-MPC reaches $93\%$ of the expert's success rate and needs \SI{18.6}{ms} per control period on a single thread, 48\% less than the expert's \SI{35.5}{ms} on sixteen parallel threads. In a controlled gap sweep, its proposed side reverses in the same gap interval as the expert's, whereas proposers without pairwise edges, trained on the same data, never reverse and all deadlock at the blocked gates of a deterministic scene.
Robotics (cs.RO)
arXiv:2610.16326 [cs.RO]
(or arXiv:2610.16326v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2610.16326 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 328 words
Related coverage
Developing Healthcare Robotics with GPU-Native Medical Physics Simulation
NVIDIA · Jul 28, 2026
Conditional Trajectory Peaks: Single-Pass Multimodal Policies over Action Chunks
arXiv · Oct 6, 2026
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · Sep 17, 2026
What's Continual Learning, and Why Might We Expect To See It In Advanced LLM Agents?
LessWrong · Jun 12, 2026