Q-attention: Enabling Efficient Learning for
Vision-based Robotic Manipulation
Stephen James Andrew J. Davison
Dyson Robotics Lab, Imperial College London
Code: https://github.com/stepjam/ARM
Trained Policies (with Q-attention visualised)
Trained Policies (with Q-attention visualised)
Videos show the Q-attention "heatmap" along with the chosen crop.
![](https://www.google.com/images/icons/product/drive-32.png)
put_rubbish_in_bin
![](https://www.google.com/images/icons/product/drive-32.png)
take_lid_off_saucepan
Keyframe Discovery
Keyframe Discovery
Visualisation of the keyframe discovery for 4 tasks.
![](https://www.google.com/images/icons/product/drive-32.png)
put_rubbish_in_bin
![](https://www.google.com/images/icons/product/drive-32.png)
reach_target
![](https://www.google.com/images/icons/product/drive-32.png)
stack_wine
![](https://www.google.com/images/icons/product/drive-32.png)
take_lid_off_saucepan
Example Demonstrations
Example Demonstrations
![](https://www.google.com/images/icons/product/drive-32.png)