arXiv:2508.12435v2 Announce Type: replace-cross Abstract: While gesture recognition using vision or robot skins is an active research area in Human-Robot Collaboration (HRC), this paper explores deep learning methods relying solely on a robot's built-in joint sensors, eliminating the need for external sensors. We evaluated various convolutional neural network (CNN) architectures and collected a dataset to study the impact of data representation and model architecture on the recognition accuracy. Our results show that spectrogram-based representations significantly improve accuracy, while model
This paper leverages advanced deep learning techniques (convolutional neural networks) and optimized data representation (spectrograms) as these methods mature, enabling more sophisticated analysis of built-in sensor data.
This development allows for more intuitive and reliable human-robot collaboration by eliminating the need for expensive and fragile external sensors, reducing complexity and cost for industrial robotic applications.
Industrial robots can now interpret human gestures through their intrinsic joint sensors, paving the way for more integrated and cost-effective human-robot interfaces across various manufacturing and operational environments.
- · Industrial robot manufacturers
- · Factories/logistics utilizing HRC
- · Deep learning algorithm developers
- · Manufacturers of external gesture sensors for robotics
- · Integrators reliant on complex external sensor setups
Wider adoption of human-robot collaboration in industrial settings due to lower entry barriers and increased robustness.
Development of new industrial safety protocols and standards tailored to robots that can interpret human intent directly from built-in sensors.
Enhanced modularity and adaptability of robotic systems, as they can be deployed and reconfigured more easily without needing recalibration of external sensing equipment.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI