Exploring Adversarial Obstacle Attacks in Search-based Path Planning for Autonomous Mobile Robots

TL;DR

The paper investigates adversarial obstacle attacks on A* path planning for autonomous robots, introducing a brute-force cost-map manipulation method to delay navigation. It combines a formal threat model, simulation in Gazebo with TurtleBot and real-world testing on a Unitree Go1, executed in parallel with normal planning. Key findings show high attack success rates and environment-dependent delays, with particularly large disruptions in constrained environments like tunnels. The work emphasizes the need for robust, attack-resilient path planning and cost-map defense mechanisms in real-world robotic deployments.

Abstract

Path planning algorithms, such as the search-based A*, are a critical component of autonomous mobile robotics, enabling robots to navigate from a starting point to a destination efficiently and safely. We investigated the resilience of the A* algorithm in the face of potential adversarial interventions known as obstacle attacks. The adversary's goal is to delay the robot's timely arrival at its destination by introducing obstacles along its original path. We developed malicious software to execute the attacks and conducted experiments to assess their impact, both in simulation using TurtleBot in Gazebo and in real-world deployment with the Unitree Go1 robot. In simulation, the attacks resulted in an average delay of 36\%, with the most significant delays occurring in scenarios where the robot was forced to take substantially longer alternative paths. In real-world experiments, the delays were even more pronounced, with all attacks successfully rerouting the robot and causing measurable disruptions. These results highlight that the algorithm's robustness is not solely an attribute of its design but is significantly influenced by the operational environment. For example, in constrained environments like tunnels, the delays were maximized due to the limited availability of alternative routes.

Figures (8)

• Figure 1: Real-world deployment without any attacks (top) and with the obstacle attack (bottom), showing the planned path in red. The attack does not change the physical environment, only the robot's cost map.
• Figure 2: Parallel execution of the obstacle attack.
• Figure 3: Navigation stack for the real robot deployment.
• Figure 4: Goal positions used in the experiments (left) and all the calculated obstacle positions (right).
• Figure 5: A comparison of robotic navigation times under benign and adversarial conditions, showing the impact of obstacles on travel time relative to Euclidean distance.