ILeSiA: Interactive Learning of Robot Situational Awareness from Camera Input

Video Materials 🔗 Paper at ArXiv 📰 PDF Results Dataset Source code

Petr Vanc (1), Giovanni Franzese (2), and Jan Kristof Behrens (1), Cosimo Della Santina (2),
Karla Stepanova (1), Jens Kober (2), and Robert Babuska (1,2)

(1) Czech Technical University in Prague, Czech Institute of Informatics, Robotics, and Cybernetics
(2) Cognitive Robotics, Delft University of Technology, The Netherlands
petr.vanc@cvut.cz, G.Franzese@tudelft.nl, jan.kristof.behrens@cvut.cz, C.DellaSantina@tudelft.nl, karla.stepanova@cvut.cz, J.Kober@tudelft.nl


Abstract

Learning from demonstration is a promising way to teach robots new skills. However, a central challenge in executing acquired skills is the ability to recognize faults and prevent failures. This is essential since the demonstrations usually cover only a limited number of mostly successful cases. During task execution, unexpected situations that were not encountered during demonstrations may occur. Examples include changes in the robot's environment or interaction with human operators. To recognize such situations, this paper focuses on teaching the robot situational awareness by using a camera input and labeling frames as safe or risky. We train a Gaussian Process regression model fed by a low-dimensional latent space representation of the input images. The model outputs a continuous risk score ranging from zero to one, quantifying the level of risk evidence at each timestep. This allows for pausing task execution in unsafe situations and directly adding new training data, labeled by the human user. Our experiments on a robotic manipulator show that our proposed method can reliably detect both known and novel faults using only a small amount of user-provided data. In contrast, a standard Multi-Layer Perceptron performs well only on faults it has encountered during training.

Video

Video showing the usage of the proposed system.

Materials

Video of skill execution

Uncut video of Open a door skill:
1. Recording demonstration,
2. Playing recorded demonstration with different Robothon box position,
3. Labeling during execution,
4. Skill trial: Detection of door closed at the start and door opened at the end. It repeats when wrong door state is detected.
For more details, see: github.com/platonics-delft/ILeSiA?tab=readme-ov-file#demo-day-with-robothon-box