MEMBER- M. Saad Khan, Dr. Mayank Baranwal (TCS Research), Barnita Maity

We propose a novel approach for multi-agent safe control, utilizing Control Lyapunov Function-based quadratic programs (CLF-QPs) to enable goal tracking while avoiding collisions. Initially designed for single-agent systems, our method effectively maintains goal tracking even in narrow passages. By incorporating collision avoidance constraints, we extend this approach to decentralized multi-agent control, ensuring scalability and computational efficiency. Extensive experiments with robots validate its effectiveness, even in highly occluded environments with up to 60 agents.

MEMBER- Maitreyee Dutta, Dr. Antonio Loria (L2S, CNRS), Dr. Elena Panteley (L2S, CNRS), Emmanuel Nuno (Univ. of Guadalajara)

We propose a solution for dynamic leaderless consensus control in networked systems affected by biased measurements. Our method redesigns a distributed consensus controller to accommodate biased data from neighbors and own state measurements. By solving a Riccati equation and designing an estimator akin to model-reference-adaptive control, we successfully compute bias estimates and ensure stability of the consensus manifold, even in the presence of constant biases.

MEMBER- Sudip Mondal

In modern lunar exploration, precise soft landing systems are essential for establishing research stations on the Moon's surface. Leveraging computer vision, particularly the Scale Invariant Feature Transform (SIFT), allows for real-time self-position determination of landing units and selection of safe landing sites. This approach enhances autonomous navigation and mission success on the lunar surface.

MEMBER-  Barnita Maity

Dissensus in intelligent traffic control diverges from the traditional goal of converging all agents to a single state or trajectory, instead emphasizing the importance of maintaining diversity among agents. This approach employs adaptive control strategies, integrating trajectory-following and collision avoidance components. Trajectory-following algorithms guide vehicles efficiently towards their destinations, while collision avoidance systems dynamically adjust trajectories to prevent accidents and ensure safety. Adaptive control enables real-time parameter adjustments based on environmental feedback, optimizing performance and safety. By prioritizing diversity, performance, and safety, anti-consensus strategies facilitate efficient and safe navigation in complex traffic environments.