So many gems in this interview, just little spoiler:

#DwarkeshPatel Next token prediction!

#RichardSutton That’s not a goal. It doesn’t change the world…

https://youtu.be/21EYKqUsPfg

ps: My goal now is aging with such clarity thinking, and relaxed dialectical teaching!!!

#TheBitterLesson […]

TubeDAgger: Reducing Expert Interventions with Stochastic Reach Tubes

TubeDAgger uses reach‑tube safety checks, so novice acts until tube leaves safe set. Tests report fewer expert interventions than DAgger, matching performance. getnews.me/tubedagger-reducing-expe... #tubedagger #imitationlearning

Labyrinth Benchmark Boosts Generalisation Testing in Imitation Learning

Labyrinth is a fully observable grid‑world benchmark that lets researchers vary maze walls, start/goal points and tasks like keys or ice, with known optimal actions. Read more: getnews.me/labyrinth-benchmark-boos... #imitationlearning #labyrinth

Soft Actor-Critic Boosts Imitation Learning for Robot Motion

Soft Actor‑Critic combined with Adversarial Motion Priors gave higher imitation rewards than PPO, improving quadruped gait on varied terrain. Read more: getnews.me/soft-actor-critic-boosts... #robotics #imitationlearning

Vehicle Trajectory Data Enhances Imitation Learning for Autonomous Cars

Nearby vehicle trajectories added to imitation-learning data cut collision rates, and using only 10 % of the original data matched full-dataset performance. Read more: getnews.me/vehicle-trajectory-data-... #imitationlearning #autonomousdriving

Ratatouille Imitation Learning Improves Real-World Robot Navigation

Ratatouille, an offline imitation‑learning system, cut collisions per meter by six‑fold and tripled navigation success in real‑world campus tests using just 11 hours of demo data. getnews.me/ratatouille-imitation-le... #robotics #imitationlearning

RoboManipBaselines: Open Framework for Imitation Learning in Robotics

RoboManipBaselines is an open-source framework for robot imitation learning that works across simulation and real hardware, released in September 2025. getnews.me/robomanipbaselines-open-... #robomanipbaselines #imitationlearning

Hybrid Imitation Learning Improves 2D Shooter RL Agent

A hybrid pipeline blends offline imitation learning with online RL, producing a 2D shooter AI that wins over 70% of matches against rule‑based foes, beating pure DQN results. Read more: getnews.me/hybrid-imitation-learnin... #imitationlearning #gamesai

Self-Augmented Trajectory Enables Imitation Learning from One Demo

SART enables robot policy learning from a single human demo by auto‑generating safe, collision‑free trajectories within annotated precision spheres. Read more: getnews.me/self-augmented-trajector... #imitationlearning #robotics #safety

The intrinsic motivation of reinforcement and imitation learning for sequential tasks YouTube video by Nguyen Sao Mai

I finally uploaded my presentation "The #IntrinsicMotivation of #ReinforcementLearning and #ImitationLearning for sequential tasks" which overviews my past 10 years of research
📽️ youtu.be/tPndN5dntoQ
#ActiveImitationLearning #CompositionalLearning #ActivityRecognition #ADL #RobotLearning #RobotCoach

How a big shift in training LLMs led to a capability explosion https://arstechni.ca #reinforcementlearning #imitationlearning #explainers #Features #AI

Work done at the QUT Centre for Robotics in collaboration with Robert Lee, @halfnelson-au.bsky.social and Niko Suenderhauf.

#robotics #imitationlearning #behaviourcloning

Neuro-Symbolic Imitation Learning: Discovering Symbolic Abstractions for Skill Learning Imitation learning is a popular method for teaching robots new behaviors. However, most existing methods focus on teaching short, isolated skills rather than long, multi-step tasks. To bridge this gap...

⏰ ICRA time: “Representation Learning 2” Session, Wednesday, starting 09:55h.

📄 Paper: arxiv.org/abs/2503.21406

🔗 Homepage: hri-eu.github.io/NeuroSymboli...

#Robotics #AI #ImitationLearning #NeuroSymbolic #ICRA2025

Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware Fine manipulation tasks, such as threading cable ties or slotting a battery, are notoriously difficult for robots because they require precision, careful coordination of contact forces, and closed-loo...

ALOHA - a low-cost bimanual robot system that learns fine manipulation tasks—like opening a condiment cup—with 80–90% success using just 10 minutes of human demos. 

Paper: arxiv.org/abs/2304.13705

Project: tonyzhaozh.github.io/aloha 

#Robotics #AI #ImitationLearning

LocoMuJoCo offers 12 humanoid + 4 quadruped envs and 22 k mocap clips for blazing‑fast imitation & RL research. Explore: github.com/robfiras/loc... #ImitationLearning #Robotics

Our STS sensor before and during contact (right column) with a cabinet knob (middle column) during a door opening task (left column). In visual mode, the camera sees through the gel membrane, allowing the knob to be found, while tactile mode provides contact-based feedback, via gel deformation and resultant dot displacement, upon initial contact and during opening. Red circles highlight the knob in the sensor view.

In a recent T-RO paper, researchers show how complex manipulation tasks that require both precise reaching and controlled slipping or sliding can benefit from see-through visuotactile sensing.
ieeexplore.ieee.org/document/108...

#RobotSensingSystems #TactileSensor #ImitationLearning

#IROS2025 reviews are coming up! Please let me know if you're interested in reviewing.

Thorough and constructive reviews are a backbone for scientific progress, please reach out!

#robotics #robotlearning #reinforcementlearning #transferlearning #imitationlearning

🚀 Exciting news! Our work has been published in IEEE 𝘙𝘰𝘣𝘰𝘵𝘪𝘤𝘴 𝘢𝘯𝘥 𝘈𝘶𝘵𝘰𝘮𝘢𝘵𝘪𝘰𝘯 𝘓𝘦𝘵𝘵𝘦𝘳𝘴 (𝘙𝘈-𝘓). 🌟

📄 𝗧𝗶𝘁𝗹𝗲: 𝘐𝘯𝘵𝘦𝘳𝘢𝘤𝘵𝘪𝘷𝘦 𝘪𝘯𝘤𝘳𝘦𝘮𝘦𝘯𝘵𝘢𝘭 𝘭𝘦𝘢𝘳𝘯𝘪𝘯𝘨 𝘰𝘧 𝘨𝘦𝘯𝘦𝘳𝘢𝘭𝘪𝘻𝘢𝘣𝘭𝘦 𝘴𝘬𝘪𝘭𝘭𝘴 𝘸𝘪𝘵𝘩 𝘭𝘰𝘤𝘢𝘭 𝘵𝘳𝘢𝘫𝘦𝘤𝘵𝘰𝘳𝘺 𝘮𝘰𝘥𝘶𝘭𝘢𝘵𝘪𝘰𝘯

Links in comments.

#robotics #research #interactiveLearning #imitationLearning #machineLearning #IEEE