I have one question about the Train DDPG Agent for Adaptive Cruise Control example in MATLAB. The speed of the lead car is sine wave using fixed-value parameters: amplitude, bias, frequency, and phase. So the agent can explore and learn the environment using the exactly same lead car speed profile (red line in the following figure) every time.
However, if the speed profile of lead car varies, for instance I have N speed profiles known in advance, the goal is to train an agent (ego car) that can adapt to all speed profiles of lead car. Is it still achievable using DDPG?