I want to know if in real-world experiments,

[nulha]

the robot determined its global position and recognized the target/goal using only a depth camera

[nulha]

the robot determined its global position and recognized the target/goal using only a depth camera

Were additional sensors such as LiDAR, IMU, GPS, or wheel odometry also used?
Additionally, did the robot use a prebuilt map or SLAM-based localization, or did it navigate autonomously in an unknown environment without any prior mapping?

[CAI23sbP]

@nulha You can see a detail of information in a SAN-FAPL project.

P.s. I think you should see this repo

[nulha]

@nulha You can see a detail of information in a SAN-FAPL project.

P.s. I think you should see this repo

Thank you very much for your helpful response.

[WzWang-Robot]

Hi Guys,

Thanks for your interests!

We leveraged computer vision BEV algorithm from https://github.com/vita-epfl/monoloco based on camera sensors for the mapping without SLAM tech. But IMU and odometry is also employed to stable the perception. We only test the navigation perforamnce on an large open space (almost 15m * 15m) without static obstacles in the real-world experiments.

If you have any concerns, please contact with us via github or emails: wang5716@purdue.edu

Thanks,
Weizheng

[nulha]

Hi Guys,

Thanks for your interests!

We leveraged computer vision BEV algorithm from https://github.com/vita-epfl/monoloco based on camera sensors for the mapping without SLAM tech. But IMU and odometry is also employed to stable the perception. We only test the navigation perforamnce on an large open space (almost 15m * 15m) without static obstacles in the real-world experiments.

If you have any concerns, please contact with us via github or emails: wang5716@purdue.edu

Thanks,
Weizheng

Thank you very much for the detailed explanation and clarification.
We really appreciate you taking the time to explain the methodology and experimental setup.
This helps us better understand your work. Thanks again!