Abstract:Autonomous Mobile Robots operating in indoor industrial environments require a localization system that is reliable and robust. While Visual Odometry (VO) can offer a reasonable estimation of the robot's state, traditional VO methods encounter challenges when confronted with dynamic objects in the scene. Alternatively, an upward-facing camera can be utilized to track the robot's movement relative to the ceiling, which represents a static and consistent space. We introduce in this paper Ceiling-DSO, a ceiling-vision system based on Direct Sparse Odometry (DSO). Unlike other ceiling-vision systems, Ceiling-DSO takes advantage of the versatile formulation of DSO, avoiding assumptions about observable shapes or landmarks on the ceiling. This approach ensures the method's applicability to various ceiling types. Since no publicly available dataset for ceiling-vision exists, we created a custom dataset in a real-world scenario and employed it to evaluate our approach. By adjusting DSO parameters, we identified the optimal fit for online pose estimation, resulting in acceptable error rates compared to ground truth. We provide in this paper a qualitative and quantitative analysis of the obtained results.
What problem does this paper attempt to address?
This paper attempts to solve the problem of reliable and robust localization of Autonomous Mobile Robots (AMRs) in indoor industrial environments. Specifically, traditional Visual Odometry (VO) methods face challenges when dealing with dynamic objects in the scene, and these dynamic objects can affect the robot's self - motion estimation. To address this issue, the paper proposes a Ceiling - based Visual System (Ceiling - DSO) based on Direct Sparse Odometry (DSO), which tracks the robot's motion relative to the ceiling through an upward - facing camera to achieve more stable and accurate localization.
### Overview of the Main Problems in the Paper
1. **Limitations of Traditional VO Methods**:
- In dynamic environments such as factories or warehouses, there are many moving objects (such as other robots, workers, etc.), and these dynamic objects can have a negative impact on the performance of VO.
- Traditional VO methods rely on feature point extraction, and the presence of dynamic objects can lead to mis - matching or loss of feature points, thus affecting the estimation accuracy.
2. **Advantages of Ceiling - based Vision**:
- The ceiling is usually a static and consistent space, not affected by ground - level dynamic objects.
- Using an upward - facing camera can avoid the influence of ground - level dynamic objects on VO estimation and provide a more stable reference frame.
3. **Limitations of Existing Ceiling - based Visual Systems**:
- Existing systems often assume that there are specific shapes or markers on the ceiling (such as circular lights, speakers, etc.), which limits the universality of the system.
- These assumptions may not hold in open warehouses or industrial spaces because the ceiling may be very high, slanted, or unevenly distributed.
### Proposed Solution
The paper introduces Ceiling - DSO, a ceiling - based visual system based on Direct Sparse Odometry (DSO), with the following characteristics:
- **No Assumption of Specific Ceiling Shapes or Markers**: Ceiling - DSO utilizes the general formula of DSO and can work on different types of ceilings without making assumptions about the specific structure of the ceiling.
- **High Adaptability**: This system is suitable for various complex ceiling environments, including high ceilings, slanted surfaces, etc.
- **Custom - made Dataset**: Due to the lack of publicly available ceiling - based visual datasets, the authors created a real - world dataset and used it to evaluate the performance of Ceiling - DSO.
### Experiments and Evaluation
To verify the effectiveness of Ceiling - DSO, the authors conducted the following experiments:
- **Experimental Platform**: A modular industrial mobile robot platform equipped with an upward - facing camera and a LiDAR sensor was used.
- **Parameter Tuning**: The parameters of DSO (such as image size, frame rate, optimization window size) were systematically adjusted, and the influence of these parameters on the trajectory estimation accuracy and calculation time was analyzed.
- **Quantitative and Qualitative Analysis**: Through comparison with the ground - truth trajectory generated by LiDAR, detailed error analysis and visualization results were provided.
### Conclusion
Through the above research, the paper demonstrates the superior performance of Ceiling - DSO in complex dynamic environments, especially when dealing with high ceilings and the absence of obvious markers. This system provides a reliable localization solution for autonomous mobile robots in indoor industrial environments.