Fig. 4: Outdoor obstacle avoidance experiments.

a Visualization of outdoor test scenarios and walking trajectories using the wearable obstacle avoidance device (WOAD) and a white cane. When approaching moving pedestrians and bikes, the WOAD facilitated much smoother movement along the yellow trajectory, while the blue trajectory corresponding to the white cane exhibited five abrupt directional changes (Please refer to the red dashed circles). b Video frame sequences and typical detection results. Video frames in the first three columns are captured by the third-person camera, while the typical detection results in the fourth column are derived from cross-modal obstacle detection module. c–e Mobility evaluation in terms of walking velocity, time, and distance. A two-sided ANOVA was used to assess significant differences between the two devices, with specific p-values provided in the figures. Box plots depict the mean (square), median (central line), the first and third quartiles (box), whiskers extending to 1.5 times the interquartile range from the first and third quartiles, respectively (n = 120, with 10 replicates for each of 12 participants). Source data are provided with paper. f Collision avoidance rate comparison. The WOAD showcases robustness across the four scenarios, with remarkable improvements of 28%, 27%, 59%, and 60%. g E2E delay of the WOAD. Across four scenarios, the WOAD achieves delay reductions of 11.6%, 13.8%, 6.3%, and 15.6%, respectively. h Power consumption in both standby and running modes. Compared to the power requirement, the WOAD in running mode saves approximate 56% power. The shadings of solid lines (mean) represent error bars (standard deviation). n = 3 replicates for each scenario.