Papers
6
Total Citations
52
H-Index
5
About
Erhard Wieser is a leading researcher in developmental robotics and cognitive architectures, whose work focuses on how robots can autonomously acquire sensory-motor skills through self-guided learning. His major contributions center on spatiotemporal learning and predictive modeling, where he has pioneered methods that allow robots to learn from minimal data—a critical challenge in real-world robotics. Wieser’s most-cited paper, “EO-MTRNN” (2020, 11 citations), introduces an evolutionary optimization approach for hyperparameters in multiple timescale recurrent neural networks, enabling more efficient learning of complex spatiotemporal patterns. His 2011 work on accelerometer-based joint orientation estimation (10 citations) provides a practical method for robotic calibration without external sensors, while his 2016 and 2018 papers on progressive sensory-motor mapping and self-verifying cognitive architectures (each with 10 citations) propose novel frameworks for bootstrapping skills through multipurpose predictors. Wieser’s research has accumulated over 50 citations, with notable achievements including the development of a scalable multi-stage learning method for reaching tasks (2017, 5 citations) and a predictive action selector that generates meaningful robot behavior from minimal training samples (2014, 6 citations). His work is particularly influential for students and researchers interested in how robots can learn like infants—through exploration, prediction, and progressive skill acquisition—offering practical algorithms that bridge neuroscience and robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2Accelerometer based robotic joint orientation estimation10 citations · 2011
- 3Progressive learning of sensory-motor maps through spatiotemporal predictors10 citations · 2016
- 4
- 5
- 6A scalable method for multi-stage developmental learning for reaching5 citations · 2017