In this work, a new robust controller is designed and analysed providing an autonomous vehicle, moving within a 2D - plain, with the ability to avoid collision with a set of obstacles despite of the presence of uncertainties in the Autonomous Vehicle (AV) with nonholonomic dynamic. The state variables (2 plain coordinates and 3 angles) and their velocities are assumed to be measurable. The controller design is based on Integral Sliding Mode (ISM) concept, aimed to minimize a given convex (not obligatory strongly convex) function of the current state. The subgradient of this cost function is also supposed to be measurable online. An optimization type algorithm is developed and analyzed using ideas of the Averaged Subgradient (ASG) technique. The main results consist in proving the reachability of the desired regime (non stationary analogue of sliding surface) from the beginning of the process and obtaining an explicit upper bound for the cost function decrement, that is, a functional convergence is proven and the rate of convergence is estimated, providing multiple obstacle avoidance. A numerical example depicts a good performance of the suggested hereby method.