5 papers
Robometer: Scaling General-Purpose Robotic Reward Models via Trajectory Comparisons
Anthony Liang, Yigit Korkmaz, Jiahui Zhang +14
General-purpose robot reward models are typically trained to predict absolute task progress from expert demonstrations, providing only local, frame-level supervision. While effecti…
When a Robot is More Capable than a Human: Learning from Constrained Demonstrators
Xinhu Li, Ayush Jain, Zhaojing Yang +2
Learning from demonstrations enables experts to teach robots complex tasks using interfaces such as kinesthetic teaching, joystick control, and sim-to-real transfer. However, these…
Causally Robust Reward Learning from Reason-Augmented Preference Feedback
Minjune Hwang, Yigit Korkmaz, Daniel Seita +1
Preference-based reward learning is widely used for shaping agent behavior to match a user's preference, yet its sparse binary feedback makes it especially vulnerable to causal con…
Actor-Free Continuous Control via Structurally Maximizable Q-Functions
Yigit Korkmaz, Urvi Bhuwania, Ayush Jain +1
Value-based algorithms are a cornerstone of off-policy reinforcement learning due to their simplicity and training stability. However, their use has traditionally been restricted t…
MILE: Model-based Intervention Learning
Yigit Korkmaz, Erdem Bıyık
Imitation learning techniques have been shown to be highly effective in real-world control scenarios, such as robotics. However, these approaches not only suffer from compounding e…