Max Robotics — Global Service Robot Directory + US Certification

About

Miles Macklin is a prominent researcher at the intersection of physics-based simulation and robot learning, whose work has fundamentally shaped how modern AI systems are trained for complex physical tasks. Best known for leading the development of **Isaac Gym**, NVIDIA's groundbreaking GPU-accelerated physics simulation platform (322 citations), Macklin pioneered the approach of running both simulation and neural network training entirely on GPU, dramatically accelerating reinforcement learning for robotics. This work catalyzed a paradigm shift in how the field approaches policy training at scale. His contributions extend deeply into simulation methodology, including non-smooth Newton methods for deformable multi-body dynamics, primal/dual descent formulations, and differentiable contact simulation — foundational tools that enable gradient-based optimization through complex physical interactions. His Grasp'D and Fast-Grasp'D frameworks apply these differentiable simulation techniques to dexterous multi-finger grasp synthesis, while DefGraspSim advances grasping of deformable objects relevant to surgical and industrial robotics. Macklin has also addressed the critical sim-to-real gap, developing adaptive domain randomization strategies that use real-world experience to refine simulation parameters. Collectively, his body of work — spanning GPU simulation, contact mechanics, and robot manipulation — has become essential reading for researchers pushing the boundaries of physically grounded robot learning.