Due to the substantial increase in semi-disabled and disabled individuals in our society, coupled with the scarcity of caregivers, providing meals for these patients is a crucial concern. To address this issue, robots specifically designed for meal-assistance have been created and implemented. Although many meal-assistance robots have been marketed, most of them can only implement the meal-assistance function in a simple way and do not consider the psychological feelings of the patient, and there are problems of simple meal-assistance trajectory and low anthropomorphism. To address these problems, we propose a novel robotic meal-assistance trajectory planning method that uses visual sensors instead of didactic methods to determine the patients’ mouth position, which significantly improves the success rate of robot meal-assistance. On the other hand, the method adds a transition segment thereby increasing patients’ comfort and reducing the cost of meal-assistance. Specifically, the method performs a quintic polynomial interpolation on the segments of the ready-to- fetch and reset, and linear and circular interpolation on the other segments, respectively, to obtain an efficient and highly anthropomorphic robotic food-assistance trajectory. To confirm the viability of the suggested approach, simulations and real-world experiments are executed. The results show that the meal-assistance trajectory planned by our proposed method not only has the characteristics of smooth and stable, high anthropomorphism and large meal intake, but also consider the variability and psychological comfort of individual patients, and has high versatility and usability.
|