Mathys C. du Plessis
Papers
19
Total Citations
199
H-Index
10
About
Mathys C. du Plessis is a prominent researcher in Evolutionary Robotics (ER), best known for pioneering the use of Artificial Neural Networks (ANNs) as alternative robot simulators to conventional physics-based systems. His work, spanning over a decade from 2009 to 2019, consistently challenges the assumption that physics models are the only viable foundation for robotic simulation, demonstrating that neural networks can effectively learn and predict robot behavior with comparable or superior results. Du Plessis's most influential contribution — his 2012 paper on simulating robots without conventional physics (29 citations) — established a compelling case for ANN-based simulators, a thread he continued through studies on differentially-steered robots, snake robots, hexapod locomotion, and inverted pendulum control. His innovative "concurrent development" approach, simultaneously evolving both robot controllers and simulator networks, represents a particularly elegant solution to the reality gap problem that plagues ER research. His exploration of ensemble neural networks further refined the accuracy and robustness of this paradigm. With a body of work accumulating over 159 citations, du Plessis has made a meaningful and lasting impact on how researchers conceptualize simulation in evolutionary robotics, offering computationally accessible alternatives that reduce reliance on specialized physics modeling expertise.
Research Focus
Key Achievements
Top Papers
- 1Simulating Robots Without Conventional Physics: A Neural Network Approach29 citations · 2012
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10