基于强化学习的平衡重式叉车防侧翻控制设计
Anti-Rollover Control Based on Reinforcement Learning of Counterbalanced for Klift Truncks

为提升平衡重式叉车的防侧翻能力，提出了一种基于制动力动态分配方法的叉车防倾覆策略。本文首先根据相关文献建立了三自由度叉车模型，该模型根据LTR方法将工况分为安全区域、安全边界区域、异常区域，之后结合Actor与Critic强化学习决策模型，实现叉车前轮制动力的动态分配，从而达到防侧翻控制的目的。本文提出的叉车防侧倾控制方法具有改造成本低、可实用性高等优势，适合在现有的平衡重式叉车上改装，实车测试表明，基于强化学习的防侧翻减速控制，可以有效降低车辆侧翻风险，提高叉车工作安全性。
In order to improve the anti-rollover ability of balanced forklift, an anti-rollover strategy of forklift based on dynamic distribution of braking force was proposed. In this paper, a three degree of freedom forklift model is established according to the relevant literature. The model divides the working conditions into safety area, safety boundary area and abnormal area according to LTR. Then, combined with actor and critical reinforcement learning decision model, the dynamic distribution of the front wheel braking force of the forklift is realized, so as to achieve the purpose of anti-rollover control. The anti-roll control method proposed in this paper has the advantages of low transformation cost and high practicability and is suitable for retrofitting on the existing balanced forklift. The actual vehicle test and MATLAB simulation show that the anti-roll deceleration control based on reinforcement learning can effectively reduce the risk of vehicle rollover and improve the safety of forklift.