Secure Interaction-aware Multi-robot Cooperation
Abstract
Autonomous mobile robots will provide fundamental help to humans in a variety of missions, such as area monitoring, surveillance, exploration, information gathering or construction of infrastructures, potentially in cooperation with robots from other organizations. The result of this cooperation will be a heterogeneous team of robots with different capabilities and interfaces, possibly working together with humans. Robots will have to communicate with each other and share sensitive information, which is not possible with available methods.The goal of the Secure Interaction-aware Multi-robot Cooperation (SECURE) project is to provide a holistic framework that allows mobile robots to cooperate in a safe, secure, and privacy-preserving manner. Ensuring safety, security, and privacy can often lead to conflicting objectives. For example, to ensure security and privacy each robot might modify the information it shares with the others, compromising its and other agents# safety. The multi-robot coordination framework we propose in SECURE will include several interconnected modules to ensure the security, and privacy guarantees, while preserving safety.To achieve these objectives, we will work at the intersection of control theory, robotics, and computer science, as follows. By relying on risk-aware trajectory optimization and reinforcement learning, SECURE will develop an innovative interaction-aware framework that will allow robots tonavigate near other agents, whether humans or robots, and achieve the mission goals. The proposed interaction-aware framework will be enhanced with a novel active-diagnosis strategy to detect and isolate anomalies in the network of robots and spot compromised (malicious) agents. Finally, within SECURE, we will explore how to enhance the security of the network and, simultaneously, enforce safety of the interaction aware framework by appropriately shaping the navigation constraints of the robots to account for possible delays in the communication and leveraging on robust control approaches.SECURE is a multidisciplinary project that aims at strengthening ONR#s research on robotics and autonomous systems, machine learning, and cybersecurity. SECURE is in direct relation with several research programs of Divisions 311 and 333 of the C5ISRT, such as (i) Machine learning, Reasoning and Intelligence (addressing the challenge of decentralized perception and planning for cooperative teams of autonomous agents), (ii) Cyber Security and Complex Software Systems (working on advanced modeling of cyber interactions), (iii) Computational Methods for Decision Making-Large Scale Distributed Decision Making (designing new rules for communication and decision#making), (iv) Science of Autonomy (perception-based decision-making and control in complex, unstructured and cluttered naval environments), and (v) Cooperative Autonomous Swarm Technology (CAST).SECURE will thus develop an integrated approach for autonomous navigation with safe coordination in human environments, while accounting for malicious agents in the network. We will provide open-source code of the algorithms and demonstrate their capabilitieson autonomous mobile robots, such as Micro Aerial Vehicles or mobile manipulators, available within the Cognitive Robotics department of TU Delft. With SECURE it will be possible to safely and securely deploy a team of mobile robots in an unknown environment in cooperation with other parties.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 10, 2025
- Source ID
- N629092512027
Entities
People
- Javier Alonso-Mora
Organizations
- Delft University of Technology
- Office of Naval Research
- United States Navy