Non-linear optimal control via particle resolution: Aid to the manoeuvre of a robot

This paper develops a global approach via particle resolution, to a robot optimal trajectory. The resolution is a maximum performance as a maximum likelihood of the trajectory in terms of the adjoint vector initial values of the Pontryagyn Principle. The performance measure becomes likelihood and can be interpreted as probability density by normalization. The optimal solution is reached by iteration on the adjoint vector, using Bayes´ Rule.