Patrick M. Pilarski
Papers
41
Total Citations
1,513
H-Index
15
About
Patrick M. Pilarski is a prominent researcher working at the intersection of reinforcement learning, assistive robotics, and human-machine interaction. He is perhaps best known for co-developing **Horde**, a scalable real-time architecture enabling robots to build and maintain world knowledge through unsupervised sensorimotor interaction — a landmark contribution that has garnered over 300 citations and remains influential in the field of artificial intelligence. Pilarski has made substantial strides in applying machine learning to myoelectric prosthetic limb control, demonstrating how actor-critic reinforcement learning methods can allow amputees to train adaptive artificial limbs in real time, work that has collectively drawn hundreds of citations and meaningfully shaped rehabilitation robotics. His contributions extend to continuous-action reinforcement learning, tuning-free step-size adaptation, and curiosity-driven deep reinforcement learning for gentle robotic manipulation. More recently, his research has explored embodied communication between humans and robots during physical interaction, reflecting a broadening vision of intelligent, responsive machines. Across more than a decade of high-impact publications, Pilarski has consistently bridged fundamental machine learning theory with real-world assistive technology applications, making his work deeply relevant to both AI researchers and clinicians working to improve the lives of people with physical disabilities.
Research Focus
Key Achievements
Top Papers
- 1
- 2Model-Free reinforcement learning with continuous action in practice234 citations · 2012
- 3
- 4
- 5
- 6Application of real-time machine learning to myoelectric prosthesis control62 citations · 2015
- 7
- 8Tuning-free step-size adaptation52 citations · 2012
- 9
- 10