Hierarchical Decomposed-Objective Model Predictive Control for Autonomous Casualty Extraction

In recent years, several robots have been developed and deployed to perform casualty extraction tasks. However, the majority of these are overly complex, require teleoperation via either a skilled operator or specialised device, often must be present at scene navigate safely around casualty. Instead, improving autonomy such can reduce reliance on expert operators potentially unstable communication systems, while still extracting in safe manner. There stages procedure, from navigating location emergency, approaching loading casualty, finally back medical assistance location. this paper, we propose Hierarchical Decomposed-Objective based Model Predictive Control (HiDO-MPC) method for manoeuvring We implement controller ResQbot — proof-of-concept mobile rescue robot previously capable rescuing an injured person lying ground, i.e. performing procedure. HiDO-MPC achieves desired behaviour by decomposing main objective into multiple sub-objectives with hierarchical structure. At every time step, evaluates decomposed generates optimal control decision. conducted number experiments both simulation using real evaluate proposed method’s performance, compare it baseline approaches. The results demonstrate that strategy gives significantly better than approaches terms accuracy, robustness, execution time, when applied scenarios.