This paper investigates the guidance method based on reinforcement learning (RL) for the coplanar orbital interception in a continuous low-thrust scenario. The problem is formulated into a Markov decision process (MDP) model, then a welldesigned RL algorithm, experience based deep deterministic policy gradient (EBDDPG), is proposed to solve it. By taking the advantage of prior information generated through the optimal control model, the proposed algorithm not only resolves the convergence problem of the common RL algorithm, but also successfully trains an efficient deep neural network (DNN) controller for the chaser spacecraft to generate the control sequence. Numerical simulation results show that the proposed algorithm is feasible and the trained DNN controller significantly improves the efficiency over traditional optimization methods by roughly two orders of magnitude.

Cooperative driving with vehicle-to-everything (V2X) communication is a promising technique to improve traffic safety and efficiency. Intersection collision avoidance (ICA) is a typical safety application of it. This paper analyzes reliability of ICA with cooperative manual driving at the system level. First, the reliability of an ICA system is defined as the probability of the ICA system avoiding collisions or near-misses at intersections without failure under conditions that collisions or near-misses are about to happen. Post-encroachment time is used in the expression of this definition. Then, components of the ICA system are classified into four types: hardware, software, maneuver, and V2X communication, and a reliability block diagram (RBD) is applied to reveal how these components contribute to system reliability. Five ICA system patterns with different V2X communication modes and strategy types are compared based on RBD analysis. This shows that centralized strategies are more reliable than decentralized ones for V2I communication if software reliability of these two strategies is the same. Furthermore, reliabilities of ICA components are analyzed in detail, and they are classified into two categories based on their different impact modes on the system. Finally, a numerical example shows how to test reliability of an ICA system using reliabilities of its components by Monte Carlo simulation. Results show that closer distances from vehicles to their conflict point when alerted, longer driver reaction time, and smaller vehicle deceleration rates are more likely to lead to system failure, whereas communication latency has little effect on it.