Towards Safe Path Tracking Using the Simplex Architecture

TL;DR

This work tackles safe, adaptive path tracking for mobile robots in dynamic environments by introducing a Simplex-architecture framework that blends a high-performance RL controller with a high-assurance PP controller. It provides design guidelines for ensuring stability and safety within the Simplex structure, including region-of-attraction analysis, safety constraints via barrier functions, and dwell-time switching. A concrete Simplex-based path-tracking controller is developed, implemented with TD3+BC RL for the high-performance side and Pure Pursuit for high-assurance guarantees, and evaluated in ROS2 simulations and a real-world deployment. Results show that safety is consistently maintained, with competitive tracking performance, though overly conservative switching and RoA estimates can limit exploitation of the RL controller; future work aims to refine switching, explore multiple high-assurance controllers, and broaden applicability.

Abstract

Robot navigation in complex environments necessitates controllers that are adaptive and safe. Traditional controllers like Regulated Pure Pursuit, Dynamic Window Approach, and Model-Predictive Path Integral, while reliable, struggle to adapt to dynamic conditions. Reinforcement Learning offers adaptability but lacks formal safety guarantees. To address this, we propose a path tracking controller leveraging the Simplex architecture. It combines a Reinforcement Learning controller for adaptiveness and performance with a high-assurance controller providing safety and stability. Our contribution is twofold. We firstly discuss general stability and safety considerations for designing controllers using the Simplex architecture. Secondly, we present a Simplex-based path tracking controller. Our simulation results, supported by preliminary in-field tests, demonstrate the controller's effectiveness in maintaining safety while achieving comparable performance to state-of-the-art methods.

Figures (5)

• Figure 1: Autonomous mobile robot demonstrating parcel delivery on a sidewalk of 2m width.
• Figure 2: The Simplex architecture sha2001using combines a high-performance, adaptive controller with a high-assurance, fallback controller that is switched to if the decision module detects unsafe actions.
• Figure 3: Results of reachability analysis.
• Figure 4: Comparing $\hat{C}_{RL}$ and $\hat{C}_S$ controllers on $\mathcal{P}_{square}$ track.
• Figure 5: Switch from high-performance (cf. \ref{['fig:cosine_hp_mode']}) to high-assurance mode (cf. \ref{['fig:cosine_ha_mode']}) in the Simplex controller due to relative angle ($\Theta_0 = -0.52rad$) between robot and path. The state $x_{C_p}$ is visualized as a blue point in \ref{['fig:max_distance_max']}.