Skip to content
RobotWorld
← Back to Videos

HARBOR goes from one prompt to a trained robot policy

Loading video
Loading video

HARBOR, accepted at CoRL 2026, autonomously builds a task, designs rewards, trains, tunes, and evaluates a robot policy from one prompt in a simulator, reaching a policy in about 1.5 hours.

HarborCoRLSimulationResearchRobot policiesAutomated RLRobot Learning
Category: research
Author: @softraeh
Date: 2026-09-04T00:00:00
Duration: 225.767s
Source: https://x.com/softraeh/status/2095910813139656720
Reference: https://arxiv.org/abs/2606.08610

Related Videos

CoRL 2026 paper removes undesired modes after BC training
1:17

CoRL 2026 paper removes undesired modes after BC training

@zhiwen_fan_ · 631 views · 2026-09-04
CoRLBehavior Uncloning
VLAct: representation-centric VLA pre-training
0:17

VLAct: representation-centric VLA pre-training

@yokoshaan · 814 views · 2026-09-05
VLActVLA
Workshop: making real-world RL practical
0:45

Workshop: making real-world RL practical

@yigitkkorkmaz · 838 views · 2026-08-31
RLReinforcement Learning
MessyMem: Persistent Spatial Memory for Robots
0:48

MessyMem: Persistent Spatial Memory for Robots

@anuvabanwasi · 908 views · 2026-09-16
Persistent MemoryMobile Manipulation
Lumos NexCore: End-to-End Robot Skill Platform
0:56

Lumos NexCore: End-to-End Robot Skill Platform

@LumosRobotics · 683 views · 2026-09-15
Robot SkillsRobot Learning
Odyssey-3: A Foundation World Model for Physical Agents
1:37

Odyssey-3: A Foundation World Model for Physical Agents

@odysseyml · 1.2k views · 2026-09-15
World ModelsFoundation Models

As an Amazon Associate, we earn from qualifying purchases.