Autonomous Braking and Throttle System: A Deep Reinforcement Learning Approach for Naturalistic Driving

Autonomous Braking and Throttle control is key in developing safe driving\nsystems for the future. There exists a need for autonomous vehicles to\nnegotiate a multi-agent environment while ensuring safety and comfort. A Deep\nReinforcement Learning based autonomous throttle and braking system is\npresented. For each time step, the proposed system makes a decision to apply\nthe brake or throttle. The throttle and brake are modelled as continuous action\nspace values. We demonstrate 2 scenarios where there is a need for a\nsophisticated braking and throttle system, i.e when there is a static obstacle\nin front of our agent like a car, stop sign. The second scenario consists of 2\nvehicles approaching an intersection. The policies for brake and throttle\ncontrol are learned through computer simulation using Deep deterministic policy\ngradients. The experiment shows that the system not only avoids a collision,\nbut also it ensures that there is smooth change in the values of throttle/brake\nas it gets out of the emergency situation and abides by the speed regulations,\ni.e the system resembles human driving.\n