Model predictive control balances tradeoffs for safe driving amid uncertainty
Scenario MPC with STL Specifications and Pareto-Based Feasibility Repair
Robotics
Summary
Self-driving cars must navigate safely even when other vehicles behave unpredictably. This paper introduces a method that helps autonomous cars plan their actions while considering uncertain behaviors of other drivers. The approach measures how well safety rules are met and finds the best balance between different goals when perfect safety can’t be guaranteed. The method was tested in driving scenarios, showing it can keep the vehicle safe by adjusting plans when necessary.
What this means in practice
- •For autonomous vehicle engineers: Design control systems that manage safety and operational goals despite uncertain behavior of other road users.
- •For robotics control teams: Implement multi-agent control policies that formally balance tradeoffs in complex environments with unpredictable agents.
Authors
Tianhao Wu, Yiwei Lyu
Abstract
Temporal logic is a formal language for reasoning about system behaviors over time. Signal temporal logic (STL), in particular, has been used to encode spatio-temporal requirements for control synthesis in multi-agent systems, often under the assumption that agents are cooperative and their dynamics are known. However, real-world multi-agent applications, such as autonomous driving, typically involve stochastic and uncontrollable agents. Recent work explored robust control with worst-case or probabilistic formulations, but remains limited in that it either (1) certifies strict satisfaction of STL constraints without addressing feasibility recovery, or (2) relaxes infeasible constraints with ego-centric objectives. In this paper, we propose a model predictive control (MPC) framework that treats feasibility repair as a Pareto optimization problem to explicitly characterize tradeoffs among agent objectives. We further provide a probabilistic certificate on STL violation rate to formally quantify uncertainty under stochastic and uncontrollable agents. The proposed framework is evaluated on two autonomous driving scenarios. Results show that the framework recovers feasible control with demonstrated safe behaviors.