Improvement of KinectTM sensor capabilities by fusion with laser sensing data using octree

Area of Science:

• Robotics and Sensor Technology
• Artificial Intelligence and Machine Learning

Background:

• Effective sensor fusion is crucial for improving robotic system capabilities.
• Limitations in individual sensor modalities, like the Kinect(TM) sensor's field of view, can hinder robot performance.
• Integrating data from multiple sensors is essential for comprehensive environmental perception.

Purpose of the Study:

• To present a novel sensor data fusion approach to overcome Kinect(TM) sensor limitations.
• To enhance the overall field of view and utility of the Kinect(TM) sensor for robotic applications.

Main Methods:

• A sensor fusion strategy combining laser and Kinect(TM) sensors was developed.
• Sensor data was modeled in a 3D environment using octrees and probabilistic occupancy estimation.
• A Bayesian method was employed to fuse sensor information and update the 3D octree map, accounting for measurement uncertainty.

Main Results:

• The proposed sensor fusion approach significantly increased the effective field of view of the Kinect(TM) sensor.
• The integrated sensor system provided a more robust and comprehensive 3D environmental representation.
• The enhanced perception capabilities are directly applicable to various robot tasks.

Conclusions:

• Sensor fusion of laser and Kinect(TM) data effectively mitigates individual sensor limitations.
• The Bayesian-based octree update method provides accurate and robust 3D mapping for robotics.
• This approach offers a practical solution for improving robot perception and task execution.