We are addressing the trajectory planning problem for autonomous driving on campus. A reference trajectory from the waypoints are developed using splines. When there is an obstacle, the vehicle replans its path. We sample the road at some look-ahead distance. We then generate new trajectories and evaluate their cost. At an intersection, the vehicles negotiate, giving higher priorities to human drivers.

We propose trajectory planning strategies for reconfiguration of a multi-agent formation on a Lissajous curve to address multiple objectives like repeated collision-free surveillance and guaranteed sensor coverage of the area with the ability for rogue target detection and trapping. The developed strategies are decentralized and hence scalable.

This work involves tracking a convoy using a drone. The path of the convoy is not known. A novel vector field based guidance scheme is proposed which first computes a time varying ellipse that encompasses all the targets in the convoy and then ensures convergence of the aerial agent to a trajectory that repeatedly traverses this moving ellipse.

The trajectories of a unicycle can generate a plethora of patterns of parametric curves (circles, spirals, epicyclic curves like hypotrochoids) and more. We propose a family of control inputs that gives rise to the pattern. The control inputs are continuous functions of the distance between the agent and a fixed point. We characterization of the generated trajectories and the necessary conditions for their generation. These appealing patterns find applications in exploration, coverage, land mine detection, etc.

We propose a decentralized multi-robot graph exploration strategy where each robot takes an independent decision without any robot-to-robot communication. The information exchange happens through the beacons placed at the vertices. The proposed technique guarantees finite time exploration of an unknown environment. New condition for declaring completion of exploration is proposed using a modified incidence matrix.