Complete and Near-Optimal Coverage Planning and Control in Robotic Crack Filling
V. Veeraraghavan, K. Hunte, J. Yi and K. Yu, "Complete and Near-Optimal Robotic Crack Coverage and Filling in Civil Infrastructure," in IEEE Transactions on Robotics, doi: 10.1109/TRO.2024.3392077. keywords: {Robot sensing systems;Robots;Robot kinematics;Planning;Sensors;Task analysis;Heuristic algorithms;Civil infrastructure;construction robots and automation;coverage planning;motion control},
Video : We made a Crack Filling Robot！ - Coming Soon
Surface cracks exist in many civil infrastructure such as road and bridge deck surfaces, parking lots, and building surfaces, etc. To prevent the crack growth and further deterioration, it is necessary to fill these cracks with appropriate materials on time. We present a robotic crack filling system that can effectively and efficiently inspect and deliver fluidic materials to fill all the surface cracks in a specified environment. Motion planning and cracking filling motion control are the two main tasks presented in this paper. A simultaneous robotic footprint and sensor coverage planning scheme is proposed to efficiently detect all the unknown targets or cracks with range sensors and cover the targets with the robot’s footprint in a structured environment. The proposed online Sensor-based Complete Coverage (online SCC) planning minimizes the total traveling distance of the robot, guarantees the complete sensor coverage of the whole free space, and achieves near-optimal footprint coverage of all the targets. The filling motion control is captured by a model predictive control (MPC) scheme given the mobile platform trajectory. Simulation and experimental results are presented that confirm the efficiency and effectiveness of the proposed scheme.
- We discuss the optimal coverage planning with known cracks in the previous section, and the crack coverage planning algorithm does not consider the sensor coverage. In order to solve the SIFT problem, we first consider the coverage planning with known target (crack) information. Then, we generalize the algorithm to the case with unknown target information.
- The onlineSCC algorithm is a practical extension of SCC where the robot stores and incrementally constructs the crack graph online. It scans for new cracks in W and updates the crack graph while simultaneously filling it.
| SCC Flow Cahrt | SCC Simulation Result |
|---|---|
 |
 |
| oSCC Flow Cahrt | SCC Simulation Result |
 |
 |
- Jetson Nano
- Lidar
| Crack coverage (%) | Overlap (%) | Distance (m) | Time (s) | |
|---|---|---|---|---|
| Sensor exploration | 36.5 | 0 | 34.9 | ??? |
| Footprint coverage | 100 | 44.5 | 126.4 | ??? |
onlineSCC |
100 | 18.4 | 41.5 | ??? |
SCC |
100 | 14.4 | 40.1 | ??? |
onlineSCC w/o con. |
100 | 10.5 | 38.7 | ??? |
SCC w/o con. |
100 | 11.0 | 38.9 | ??? |
| Experiment | 100 | 23.0 | 48.1 | ??? |
The authors thank ...... of Rutgers University for many helpful discussions and suggestions.