Page 43 - Read Online

Page 192 Chen et al. Intell Robot 2024;4:179-95 I http://dx.doi.org/10.20517/ir.2024.11

4.4 Advantages
Combining the presented RANSAC and the KNN-augmented ICP is particularly effective in dynamic envi-
ronments and restricted viewpoints for several reasons:

1. Robustness to Motion: The ability of RANSAC to handle outliers means that the motion of the robot does
not affect the staircase shape classification.
2. Accuracy in Real-Time Environments: The efficiency of KNN in the ICP algorithm ensures that the point
cloudregistrationonlyusesthemostrelevantpointsandremainsaccurateevenwhentheenvironmentchanges,
which is crucial for real-time locomotion assistance.
3. AdaptabilitytoDifferentViewpoints: ByaugmentingICPwithKNN,thealignmentofthepointclouddatais
based on the most relevant and geometrically consistent points, which is critical in scenarios with limited field
of view. This ensures that each data segment contributes to a comprehensive understanding of the staircase
geometry.

4.5 Limitations
Itisessentialtoacknowledgecertainlimitations. Despitesignificantadvancements,thealgorithmperformance
mightstillbeinfluencedbycertainhighlyirregularoruncommonstaircasedesignsnotextensivelyrepresented
in the presented staircase point cloud classifications in Figure 1. Overcoming these limitations might necessi-
tate further refinement and expansion of the environment point cloud dataset to encompass a wider array of
staircase variations.

While improvements have been made, the method may still face challenges in extremely dynamic or unpre-
dictable environments, where rapid changes in the robot’s movement or environmental conditions could affect
the accuracy of feature extraction and point cloud registration.

4.6 Future research directions
Future research directions for this work include:

1. Algorithm Optimization for Diverse Hardware: Developing more efficient algorithms that can be effectively
implemented on a wider range of walking-aid robots with varying computational capabilities.
2. Enhanced Adaptability in Dynamic Environments: Focusing on improving the robustness of the feature
extraction method to better handle highly dynamic environments and unpredictable scenarios.
3. Integration with Advanced Sensing Technologies: Exploring the integration of emerging sensing technolo-
gies to further refine environmental perception and feature extraction accuracy.
4. Real-World Testing and Validation: Conducting extensive real-world testing to validate and refine the pro-
posed method under various environmental conditions and scenarios.

5. CONCLUSIONS
In conclusion, this paper presents significant advancements in the domain of staircase shape feature extraction
for walking-aid robots. The proposed approach successfully addresses the limitations of previous methods,
offering a more robust and accurate system for environmental perception in complex terrains, particularly
staircases. The integration of algorithms, such as RANSAC and KNN-augmented ICP, has been demonstrated
to substantially enhance the point cloud registration process, thereby improving the accuracy and efficiency of
feature extraction. The results show that the absolute trajectory error across all trials falls within the centimeter
range.

The findings demonstrate the potential of this method in enhancing the navigational capabilities of walking-
aid robots, contributing to safer and more reliable mobility assistance. However, there are still limitations,