Data & collection
Trajectory
A trajectory is a time-ordered sequence of states or observations, actions and, where applicable, rewards generated as an agent or robot evolves. A complete episode or policy rollout often yields a trajectory, but the terms are not universally identical: trajectories may be partial, while episodes have dataset- or environment-defined boundaries.
Also known as: episode, rollout
Updated
What the sequence represents
In robot-learning data, a trajectory preserves temporal dependence. A step may contain an observation, the action applied from that observation, the resulting reward and boundary flags. The next step reflects the consequences of the previous action. Images, joint state, force readings and commands therefore need timestamps and an explicit alignment convention.
The exact fields are format-specific. RLDS stores datasets as episodes containing ordered steps, with flags for the first, last and terminal steps. DROID uses trajectories to package real robot interaction with camera, state, action and task information. In motion planning and control, “trajectory” can also mean a time-parameterised sequence of desired configurations, as explained by Modern Robotics, rather than a logged learning sample.
Trajectory, episode and rollout
These terms often refer to the same stored sequence, but their emphasis differs. An episode is an interaction bounded by a reset, termination, timeout or collection rule. A rollout is the act, or result, of executing a policy or controller for some horizon. The generated observations and actions form a trajectory.
They should not be treated as universal synonyms in data specifications. A dataset may save a short window from an episode as one trajectory, concatenate several skill segments, or call a truncated policy execution a rollout. For humanoid data, boundaries affect returns, success labels and sequence-model context. Consumers should check the dataset’s own terminal flags, reset rules, sampling rate and definition of a step before combining trajectories.
Sources
Related terms
Data & collection
Demonstration
A demonstration is a recorded example of how an intended task or behaviour is performed, usually represented as a time-aligned sequence of observations, states and actions. For humanoid robot learning, demonstrations may come from teleoperation, kinaesthetic guidance, motion capture or autonomous experts and provide targets for imitation.
Models & learning
Policy
A policy is the decision rule that maps a robot’s current observations or estimated state, and sometimes a task instruction, to an action or probability distribution over actions. It can be hand-designed or learned from demonstrations, rewards or both. In humanoid robotics, its outputs may be joint targets, torques, end-effector changes or higher-level skills.
Models & learning
Reinforcement learning
Reinforcement learning is a method in which an agent learns a policy by interacting with an environment and optimising cumulative reward. In humanoid robotics, actions change the robot and world, while observations, rewards and episode endings provide experience for improving balance, locomotion or manipulation behaviour.
Hardware & control
Proprioception
Proprioception is sensing of a robot’s own internal configuration and motion rather than the external scene. For a humanoid it commonly includes joint positions and velocities, actuator effort or torque, and inertial measurements of body rotation and acceleration. These signals support state estimation and feedback control but do not, by themselves, directly describe nearby objects or terrain.