Author: Sayantani Bhattacharya
In times of disaster, every second counts, and reaching survivors in hazardous terrains poses significant challenges. Imagine a coordinated team of agile, four-legged robots, working together to navigate treacherous environments like dense forests or mines. These quadrupedal robots autonomously perform simultaneous localization and mapping (SLAM), creating real-time detailed maps of their surroundings. By employing decentralized collaborative system, these robots can share and merge their individual maps, creating a comprehensive understanding of the complete area without relying on a central system. This approach enhances the robustness and speed of search operations, as the failure of a single unit does not compromise the entire mission. Quadrupeds inherently work well in uneven terrains, and harnessing the strengths of SLAM to explore unmapped areas with LIDAR and Visual-Inertial sensor data, these robotic swarms represent a leap forward in disaster response, offering hope and assistance when it’s needed most. By all means this is just the first iteration and needs good work for being deployable onsite.
For more details about the project please refer my Portfolio Post
- Complete system:
-
System Architecture of GO2:
-
System Architecture of GO1:
- Unitree GO1 & GO2.
- Zed 2i Camera & On-board 4D Lidar.
- Jetson Orin Nano.
- buck-convertor (24V->12V).
- 3D print the mount for unitree.
- Display port adapters for Jetson.
- Ethernet cable for initial testing with unitree sdk.
- Micro SD cards.
- C++
- ROS2 Jazzy and Humble
- Python
- Unitree SDK - GO1 and GO2
- Zed SDK
- Slam, RTabMap, and Nav2 pkg