Method and system for determining road surface friction of an autonomous vehicle using learning based model predictive control

摘要

In one embodiment, planning data is received from, for example, a planning module to travel an autonomous vehicle (ADV) from a departure location to a destination location. In response, a series of control instructions are generated based on the planning data, wherein the control instructions are applied at different times from the location of departure to the location of destination. The cost function is calculated by applying the cost function to the control command, the first road surface friction estimated on the current stroke and the second road surface friction estimated between the previous travel from the departure position to the destination position, and the cost is calculated. Ru. Using the cost function taking into account the end cost before the previous stroke, the first road surface friction of the current stroke is estimated to minimize the cost. [Selected figure] Figure 6