Computer Vision-Based Adaptive Semi-Autonomous Control of an Upper Limb Exoskeleton for Individuals with Tetraplegia

We propose the use of computer vision for adaptive semi-autonomous control an upper limb exoskeleton assisting users with severe tetraplegia to increase independence and quality life. A tongue-based interface was used together such that individuals complete were able it despite being paralyzed from neck down. The uses detect nearby objects estimate how grasp them assist user in controlling exoskeleton. Three schemes tested: non-autonomous (i.e., manual using tongue) control, a fixed level autonomy, confidence-based autonomy. Studies on experimental participants without carried out. evaluated both terms their performance, as time number commands needed given task, well ratings users. studies showed clear significant improvement performance when either schemes. outperformed version some scenarios, namely, more complex tasks training system.