Synergistic Perception and Control Simplex for Verifiable Safe Vertical Landing

TL;DR

This paper tackles the challenge of verifiably safe vertical landings for autonomous air taxis in cluttered environments by integrating perception and control through a Perception Simplex that is enhanced with Dynamic Confirmation. It replaces static worst-case control assumptions with real-time confirmation of capabilities via an $\mathcal{L}_1$ adaptive controller and a sliding-window estimate of $a_{max}$, enabling faster landings without compromising safety. The approach combines a verifiable safety layer with a mission layer, using a safety override to prevent collisions and a low-latency, robust control loop to follow safe braking trajectories. Evaluations in a software-in-the-loop setting (CARLA + GUAM) demonstrate substantial reductions in landing time (e.g., $34.2\%$) while maintaining zero collisions across scenarios, highlighting practical improvements for verifiable autonomous aerial mobility.

Abstract

Perception, Planning, and Control form the essential components of autonomy in advanced air mobility. This work advances the holistic integration of these components to enhance the performance and robustness of the complete cyber-physical system. We adapt Perception Simplex, a system for verifiable collision avoidance amidst obstacle detection faults, to the vertical landing maneuver for autonomous air mobility vehicles. We improve upon this system by replacing static assumptions of control capabilities with dynamic confirmation, i.e., real-time confirmation of control limitations of the system, ensuring reliable fulfillment of safety maneuvers and overrides, without dependence on overly pessimistic assumptions. Parameters defining control system capabilities and limitations, e.g., maximum deceleration, are continuously tracked within the system and used to make safety-critical decisions. We apply these techniques to propose a verifiable collision avoidance solution for autonomous aerial mobility vehicles operating in cluttered and potentially unsafe environments.

Figures (13)

• Figure 1: Proposed system design for autonomous air taxis. The mission layer represents the complex, high-performance, although unverifiable, autonomy software. The safety layer provides verifiable collision avoidance.
• Figure 2: LiDAR sensor on the air taxi, marking the vertical landing path as a green area.
• Figure 3: Obstacle size limit for detectability guarantee. The green shaded area signifies the conditions where the obstacle is guaranteed to be detected, given other constraints are also met ($\S$\ref{['sec:model_detectability']}). Note that this is a pessimistic model, i.e., obstacles may still be detected in some parts of the unshaded area of the plot.
• Figure 4: Projection of the obstacle to a plane towards the air taxi (origin), at the obstacle's point closest to the air taxi.
• Figure 5: The framework of the control module comprises a baseline controller, an $\mathcal{L}_1$ adaptive controller, and the estimation of maximum acceleration $a_{max}$.