Exploring groundbreaking research across Fundamental Networks, Racing, Reinforcement Learning, and Robotics
Discover papers across four key research domains at the forefront of artificial intelligence and robotics
Core architectures and innovations in neural networks, transformers, and generative models.
0 papers View papersAlgorithms and approaches for high-speed decision making and autonomous racing systems.
3 papers View papersAdvanced RL techniques, policy optimization, and agent-environment interaction models.
0 papers View papersControl systems, embodied AI, manipulation tasks, and physical interaction learning.
0 papers View papersIn-depth reviews of influential publications making an impact on the AI research landscape
The transformer architecture introduced in this paper has revolutionized natural language processing and beyond, enabling efficient parallel computation and capturing long-range dependencies.
Read reviewSuper-Human Performance in Gran Turismo Sport generalizes the approach to autonomous racing, achieving superhuman performance in high-fidelity racing simulations using a single deep reinforcement learning algorithm with no prior knowledge beyond sensor-level inputs and game physics.
Read reviewTrajectory-aided Learning generalizes deep reinforcement learning for autonomous racing, achieving high-speed performance using only LiDAR input by incorporating classical trajectory planning into the reward formulation—without requiring any explicit localization or mapping.
Read reviewThis work demonstrates how reinforcement learning enables the Shadow Dexterous Hand to manipulate physical objects with unprecedented dexterity, solving the Rubik's Cube with a single robotic hand.
Read review