Vision based obstacle crossing for quadruped robot
In recent years, quadruped robots have gained significant attention due to their ability to traverse complex terrains. However, most researches use a combination of LiDAR and other sensors to identify obstacles, which not only increases the production costs of robots but also makes the overall struc...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Master by Coursework |
| Language: | English |
| Published: |
Nanyang Technological University
2025
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/182223 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | In recent years, quadruped robots have gained significant attention due to their ability to traverse complex terrains. However, most researches use a combination of LiDAR and other sensors to identify obstacles, which not only increases the production costs of robots but also makes the overall structural design more complex. Thus, this dissertation introduces an obstacle-crossing method for quadruped robots based purely on vision. This study can be divided into two parts. The first part focuses on the exploration of motion control algorithms for quadruped robots, including posture adjustments and walking. The second part is dedicated to the study of visual perception algorithms. The experiment uses a stereo depth camera to generate point cloud data, combined with RGB images for multimodal fusion. This approach is employed to classify obstacles and select appropriate methods for crossing them. The results show that the model trained using this method has a high accuracy in distinguishing obstacle sizes. All the experiments are conducted in the Gazebo simulation environment. |
|---|