Constrained PSO-Splines Trajectory Generation for an Indoor Nanodrone

The trajectory generation problem is still challenging as it involves determining a smooth and feasible path considering convoluted constraints. This paper builds upon the strengths of fast computation of PSO (Particle Swarm Optimization) and the B-splines properties to solve an optimization problem for trajectory planning. The precomputed trajectory of an indoor quadcopter is parameterized by B-splines functions allowing to enforce operation constraints (i.e., position, velocity, angles, thrust, angular velocity, way-point passing) while minimizing the input effort. Simulation and experimental testing with a nanodrone validate our approach.