PBDM: Peak-Based Differential Method for Improve Leg Detection Performance with Laser Range Finder

Determining human presence is crucial for robots engaged in complex tasks such as avoiding and tracking obstacles. Incorporating diverse sensors in robots poses challenges and opportunities to enhance detection algorithms, which rely on the diversity of input signals for recognition. Modern methods for differentiating lower limbs use laser rangefinder signals, segmenting regions believed to be legs, and identifying distinctive features based on geometric and spatial properties of the data. This study introduces the Peak-Based Differential Method (PBDM) to improve the accuracy and efficiency of segmentation by identifying potential pins through analysis of signal peaks in one-dimensional (1D) space. Then, five specific features of human legs are described and extracted from these segments in two-dimensional (2D) signal space for object classification. The implementation of this innovative method aims to optimize the overall accuracy of the detection process, reduce the number of features needed for accurate identification, and reduce processing time. Implementation of these methodological improvements has shown promising results when applied in real-life scenarios, highlighting their potential effectiveness for complex robotics applications.