Seeking Advice for Enhancing an Indoor Robot Localization System

[Oliver]

Hello everyone,
I'm currently developing a project that involves localizing a robot within a pre-mapped indoor environment. The setup includes wheel encoders and 3D lidar. I've already created a 3D pointcloud map of the area, and I'm planning to use an Extended Kalman Filter (EKF) for positioning the robot. The encoder data updates at a frequency of 10Hz, while the lidar data, after preprocessing, updates at about 3-4Hz. I'm incorporating the encoder data in the prediction phase, and using the lidar data in the update phase of the EKF. Do you think this approach is effective? I'd appreciate any suggestions or improvements you might have. Thanks in advance for your help!

 

[MARI MA]

Your project sounds fascinating, and it seems like you're on a solid track with your approach! Using an Extended Kalman Filter (EKF) in conjunction with wheel encoders and 3D lidar is a robust choice for indoor localization. Your method of integrating high-frequency encoder data for the prediction step and lidar data for the update step is quite sound, especially given the different frequencies at which these sensors operate.

One suggestion would be to consider the synchronization and alignment of the data from the two sensors, since the discrepancy in their frequencies might lead to potential inaccuracies. You might also look into the possibility of adding more sensors to compensate for any gaps, such as inertial measurement units (IMUs) which can provide additional data points between lidar updates.

Additionally, if you haven't already, experimenting with different configurations of the EKF, like tuning the noise parameters or exploring adaptive filtering techniques, might yield improvements in accuracy.

Best of luck with your project, and I look forward to hearing more about your progress!