Abstract. We study the kinematics of the trident snake robot equipped with either active joints and passive wheels or passive joints and active wheels. A control system representation of the kinematics is derived, and control singularities examined. Two motion planning problems are addressed, corresponding to diverse ways of controlling the robot, and solved by means of the endogenous configuration space approach. The constraints imposed by the presence of control singularities are handled using the imbalanced Jacobian algorithm assisted by an auxiliary feedback. Performance of the motion planning algorithms is demonstrated by computer simulations.

The trident snake robot is a mechanical device that serves as a demanding testbed for motion planning and control algorithms of constrained non-holonomic systems. This paper provides the equations of motion and addresses the motion planning problem of the trident snake with dynamics, equipped with either active joints (undulatory locomotion) or active wheels (wheeled locomotion). Thanks to a partial feedback linearization of the dynamics model, the motion planning problem basically reduces to a constrained kinematic motion planning. Two kinds of constraints have been taken into account, ensuring the regularity of the feedback and the collision avoidance between the robot's arms and body. Following the guidelines of the endogenous configuration space approach, two Jacobian motion planning algorithms have been designed: the singularity robust Jacobian algorithm and the