Impressive work!! I wonder if there are some videos about real robots on more complex terrains?
@weili31884 жыл бұрын
May I ask that what is this mobile robot? Or you build this on your own?
@pratikprajapati86205 жыл бұрын
You are amazing hatzzzzz. offff
@gondala5 жыл бұрын
I am curious if this robot also using "virtual" sensors i.e. it is very useful in real world?
@dhayaaraissankhudher57436 жыл бұрын
Very impressive. What is the simulator?
@alexanderwinkler68216 жыл бұрын
The first clips are visualization using ros-rviz and wiki.ros.org/xpp, the simulations are done in Gazebo.
@dhayaaraissankhudher57436 жыл бұрын
Many thanks indeed!
@hyd40876 жыл бұрын
Nice work!
@jamieh60846 жыл бұрын
Was this trained using simulated imperfections in the aforementioned areas (force tracking and sensor data?) If not I would be interesting to see if it copes better in reality.
@nezv716 жыл бұрын
I think "trained" is the wrong word here. The gait trajectory generation is posed as a nonlinear programming problem and solved iteratively. A new problem is solved for every plan. The plans are then tracked by a feedback controller to help with disturbances / imperfections.
@jamieh60846 жыл бұрын
Thank you for clearing that up for me and explaining the iterative nature. I was thinking more along the neural network path, so I thought it was being "trained". Very interesting
@xiaohanfei45786 жыл бұрын
Great work! We actually have something similar to this in cvpr17.
@carlosmacedo47926 жыл бұрын
Useful!
@jamieh60847 жыл бұрын
What about combining with tango AR unit? More robust inside out tracking, for when it goes mobile.
@jamieh60847 жыл бұрын
Does this work with 3D co-ordinates on a surface? Or is it limited to 2D. Thank you
@ADRLabETH7 жыл бұрын
The approach would generally allow 3D by introducing additional optimization variables (e.g. body height and orientation) and the foothold height must be adapted to height of the terrain at that xy-position. But this is an idea for future work currently not implemented.
@marknabil99787 жыл бұрын
I didn't read the paper yet, but I have a small question. you don't need to know the location of the marker predefined to run your algorithm, Am i correct ? . It looks impressive and promising anyways, thx for sharing it.
@Usman-zp6ig8 жыл бұрын
Hi. I am curious to know which transcription method you used in particular and your experience when solving the NLP by an interior point method. I am currently using the SNOPT implementation of sqp to solve my discretized optimal control problem. Although the solver appears to find a local minimum, I am unable to determine (using the output of SNOPT) whether the local minimum is indeed a stable solution. I would be grateful to know if you've undertaken such analyses. Many thanks. Usman
@yasbeel67369 жыл бұрын
Hi I saw your video it is quite impressive and I'm interesting. could you please send me your model and code. If you agree see my email is [email protected] I'm looking forward to hear from you Regards Yaser