Design and Optimization of an Assistive Cane with Visual Odometry for Blind People to Detect Obstacles with Hollow Section
Jingyu Tang,Mingze Sun,Lingjun Zhu,Menghan Hu,Mei Zhou,Jian Zhang,Qingli Li,Guangtao Zhai
DOI: https://doi.org/10.1109/jsen.2021.3115854
IF: 4.3
2021-01-01
IEEE Sensors Journal
Abstract:Existing obstacle avoidance systems do not consider the situation of hollowed-out obstacles. Therefore, in this paper, we design an assistive cane with visual odometry based on these requirements of the blind to aid them in attaining safe indoor navigation. The proposed device is portable, compact and adaptable. The assistive cane with visual odometry can be used not only to directly detect obstacles, but also to draw the attention of people around, providing multiple protection for the blind. In order to optimize the device parameters, we select the height of the camera and the speed of the voice broadcast as factors to design a complete factorial design, aiming to explore the parameter level that can make the passing time shortest and the success rate of obstacle avoidance highest. We invited two visually impaired people and two blindfolded students with normal vision as the experimental subjects. Each subject was required to conduct the experiments under nine combinations of device parameters. The subjects tried to avoid obstacles according to the voice prompts. The passing time and the success rate of obstacle avoidance were recorded. The experimental results show that the broadcast speed causes the significant effect to the success rate and both the two factors affect the passing time. In addition, we discuss the results of different subjects’ response surfaces and perceptual patterns and get the conclusion that different visual perceptual patterns, namely route representation and two different survey representations, have a certain impact on the setting of the optimal parameter level and the optimal passing time. The congenitally blind subject with route representation has the shortest optimal passing time in an unfamiliar environment, and the camera height should be set at the center of the hollowed-out obstacle, while the results are somewhat different for the acquired visual impairment subjects and the blindfolded normal vision subjects with survey representation. The camera height should be set between the center and the upper position of the obstacle. In summary, the camera height should be set at the center of the hollowed-out barrier and the broadcast speed should be set at the highest level. At the same time, the device parameters should be adjusted appropriately according to the visual perception pattern of users. Further researches are needed to analyze the perceptual patterns in depth.