Model predictive control (MPC) has been widely researched for automotive control. However, the real-world application of MPC for autonomous vehicles (AV) is still limited due to the high computational requirement of solving the real-time optimization problem. To address this issue, this paper presents an event-triggered MPC framework and illustrates its application in AV path tracking control using the CARLA simulation environment. Unlike traditional time-triggered MPC, event-triggered MPC solves the optimization problem only when an event is triggered. Otherwise, the previously optimized control sequence is used to determine control action. Therefore, when following the same path, event-triggered MPC solves fewer optimization problems than time-triggered MPC, thereby reducing the computation requirements. To demonstrate the effectiveness of the proposed approach, both time-triggered MPC and event-triggered MPC are simulated and compared using CARLA. Results validate that the proposed event-triggered MPC method reduces significant computation with acceptable performance degradation.

Event-triggered model predictive control (eMPC) is a popular optimal control method with an aim to alleviate the computation and/or communication burden of MPC. However, it generally requires a priori knowledge of the closed-loop system behavior along with the communication characteristics for designing the event-trigger policy. This paper attempts to solve this challenge by proposing an efficient eMPC framework and demonstrates successful implementation of this framework on the autonomous vehicle path following. First of all, a model-free reinforcement learning (RL) agent is used to learn the optimal event-trigger policy without the need for a complete dynamical system and communication knowledge in this framework. Furthermore, techniques including prioritized experience replay (PER) buffer and long short-term memory (LSTM) are employed to foster exploration and improve training efficiency. In this paper, we use the proposed framework with three deep RL algorithms, i.e., Double Q-learning (DDQN), Proximal Policy Optimization (PPO), and Soft Actor-Critic (SAC), to solve this problem. Results show that all three deep RL-based eMPC (deep-RL-eMPC) can achieve better evaluation performance than the conventional threshold-based and previous linear Q-based approach in the autonomous path following. In particular, PPO-eMPC with LSTM and DDQN-eMPC with PER and LSTM obtain a superior balance between the closed-loop control performance and event-trigger frequency. The associated code is open-sourced and available at: https://github.com/DangFengying/RL-based-event-triggered-MPC .

We study the prognosis of fault, i.e., its prediction prior to its occurrence, in stochastic discrete event systems. We introduce the notion of m-steps Stochastic-Prognosability, called S _m -Prognosability, which allows the prediction of a fault at least m-steps in advance. We formalize the notion of a prognoser and also show that S _m -Prognosability is necessary and sufficient for the existence of a prognoser that can predict a fault at least m-steps prior to occurrence, while achieving any arbitrary false alarm and missed detection rates. We also provide a polynomial algorithm for the verification of S _m -Prognosability. Finally, we compare the notion of stochastic prognosability with that of stochastic diagnosability, and show that the former is a stronger notion, as can be expected.