Answers are grounded in this paper and cite their sources.
Robot Operating System 2 (ROS 2)-based robotic systems face a distinctive security challenge: attacks that pass network-level checks can still violate the physical laws governing motion, sensing, and control. Physics-based detection addresses this gap by monitoring kinematic and dynamic invariants, including motor-level signals such as power balance and current–velocity correlations recently demonstrated on embedded controllers operating under real-time constraints. While such methods are well established in industrial cyber-physical systems, their application to the ROS 2 middleware stack and the heterogeneous mobile platforms it supports remains largely unexamined. This survey synthesizes research across industrial electronics, robotics, cybersecurity, and control theory through a structured review of 183 papers. The principal contributions are a unified attack taxonomy mapping each vector to its physics-based detectability, an analysis of how the decentralized data distribution service (DDS) architecture of ROS 2 complicates unified physics monitoring on resource-constrained hardware, and a prioritized research roadmap. Empirical case studies spanning control-invariant detection on embedded platforms, convolutional-neural-network-based motion recognition, and a ROS 2-native intrusion prevention system are comparatively evaluated. Physics-based validation emerges as a promising complement to conventional cyber defenses, though evidence from native ROS 2 deployments remains limited, and middleware-level integration of physics monitors on resource-constrained embedded platforms is identified as the key unresolved challenge.
No takes yet. Share an insight, caveat, or question.
Silva et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: