Glossary
The terms that decide whether a policy runs on your robot. Every entry cites where the idea comes from.
- Action chunking · chunked action prediction, ACT-style chunking
- Predicting a sequence of future actions per model call and executing them before querying the policy again, instead of predicting one action per control step.
- Action horizon · prediction horizon, execution horizon, chunk size, n_action_steps
- How many future action steps a policy predicts per call (prediction horizon) and how many it executes before replanning (execution horizon).
- Action tokenization · action discretization, action tokens
- Converting continuous robot actions into discrete tokens so that an autoregressive transformer can predict them the way it predicts text.
- Affordance · action possibility, affordance map
- What actions an object or scene makes possible for a particular agent; in robot learning, a learned estimate of which actions are feasible and useful in the current state.
- Asynchronous inference · async inference, PolicyServer / RobotClient
- Running policy inference in parallel with action execution, so the robot keeps acting from an action queue while the next chunk is computed.
- Behavioral cloning · BC, behavior cloning
- Supervised learning of a policy that maps observations to the actions a demonstrator took, with no reward signal or environment interaction during training.
- Control rate · control frequency, control loop rate, control_rate_hz
- How many actions per second the robot's controller executes, in Hz; set by the robot and dataset, and distinct from how often the policy model is called.
- DAgger · Dataset Aggregation
- Dataset Aggregation: an iterative imitation-learning algorithm that runs the learned policy, has an expert label the states it visits, and retrains on the combined data.
- Degrees of freedom · DoF, DOF
- The number of independent coordinates needed to specify a mechanism's configuration; for a serial robot arm, usually its number of actuated joints.
- Diffusion policy · Diffusion Policy, action diffusion
- A policy that generates action sequences by iteratively denoising random noise with a conditional diffusion model, introduced by Chi et al. in 2023.
- Embodiment gap · cross-embodiment gap, embodiment mismatch
- The differences in kinematics, action space, sensors and dynamics between the robots a policy was trained on and the robot it is deployed on.
- Flow matching · conditional flow matching, CFM
- A generative training method that regresses a velocity field carrying noise to data; VLAs such as π0 and SmolVLA use it to generate continuous action chunks.
- Human-in-the-loop reinforcement learning · HIL-RL, HIL-SERL, human-in-the-loop RL
- Real-robot reinforcement learning in which a human supplies demonstrations and takes over during training to correct the policy, making on-hardware RL practical.
- Imitation learning · IL, learning from demonstration, LfD
- Learning a policy from demonstrations of the desired behavior rather than from a hand-written controller or a reward signal.
- Inference latency · policy latency, inference time, inference_latency_ms
- Wall-clock time for one policy call from observation to action chunk. Not the same as control rate: action chunking decouples the two.
- Leader–follower arm · leader arm, follower arm, puppeteering, joint-space teleoperation
- A teleoperation setup in which a human moves a passive 'leader' arm and a kinematically matched 'follower' robot copies its joint positions in real time.
- LeRobotDataset · LeRobotDataset v3.0, LeRobot dataset format
- Hugging Face LeRobot's standard format for robot demonstration data: Parquet for states and actions, MP4 for camera streams, plus episode metadata.
- Parameter-efficient fine-tuning (PEFT) and LoRA · PEFT, LoRA, Low-Rank Adaptation
- Adapting a pretrained model by training a small set of added parameters while the original weights stay frozen; LoRA learns low-rank update matrices.
- Proprioception · proprioceptive state, robot state, observation.state
- The robot's measurement of its own body state, such as joint positions, velocities and gripper opening, given to a policy alongside camera images.
- Quantization and numeric precision · fp32, bf16, fp16, int8, int4, reduced precision
- Running a model's weights or activations at lower numeric precision (bf16, fp16, int8, int4) to save memory and often time; every latency number must state it.
- Real-time chunking · RTC
- An inference-time method that generates the next action chunk while the current one executes, freezing actions that will run anyway and inpainting the rest.
- Reward model · learned reward, reward classifier, success detector
- A learned function that scores how well a trajectory or state accomplishes a task, used in place of a hand-written reward for RL, data weighting or evaluation.
- Sim-to-real transfer · sim2real, reality gap
- Training or validating a policy in simulation and deploying it on physical hardware; the gap between simulated and real physics and sensing is the central obstacle.
- Teleoperation · teleop, remote operation
- Real-time human control of a robot through an input device; in robot learning, the standard way to collect demonstrations for imitation learning.
- Vision-language-action model · VLA, VLA model
- A robot policy built on a pretrained vision-language model that maps camera images, a language instruction and robot state to actions.
- World model · dynamics model, learned simulator
- A learned model that predicts how an environment's state or observations evolve in response to actions, used for planning, imagined rollouts or auxiliary training.