Chengtao Yao, Chengcheng Zhao, Peng Cheng, and Jiming Chen, Zhejiang University
Industrial robotic arms are central to modern manufacturing, with broad deployment in critical domains. Motion is a primary security concern, as it is a fundamental capability of robotic arms, and adversarial manipulation (e.g., altering production logic, positioning, or dynamics) can lead to product defects or physical damage. Remote attestation is a promising mechanism for verifying execution integrity. However, existing approaches focus on control-flow or data-flow properties and fail to capture motion semantics, limiting their ability to adequately verify the physical execution of robotic arms.
This paper presents Trajectory Integrity (TI) as a new security property that ensures a robotic arm's motion conforms to its intended path. To enforce TI, we design TAT, a minimally invasive attestation framework that leverages a Timed Motion Event Graph to capture motion semantics and combines event and joint measurements to verify actual motion. We implement a hardware-software prototype of TAT on an open-source robotic arm platform. Evaluation on real-world task programs shows that TAT incurs at most 2.30% memory overhead and 0.14% execution time overhead, demonstrating its performance and practicality. Furthermore, its attestation capability is evaluated under diverse motion-related parameter modifications, confirming its effectiveness in trajectory integrity attestation.