Awesome Reliable Robotics 🤖

A curated collection of robotics papers focused on real-world reliability and robustness. Originally a personal reference, I'm sharing this list in hopes it helps others.

Prerequisite: must include real-world results.

Contributions are welcome!

---

Name	Date	Categories	Real World Success Rate	Code	Paper	Project	Organization(s)	Notes
Contact-Anchored Policies: Contact Conditioning Creates Strong Robot Utility Models	02/2026	Representation Learning	Average zero-shot success with Picking, Opening, and Closing tasks across 4 robot arms shown below.	Code	Paper	Project	NYU, UC Berkeley, UCLA, Hello Robot, Ai2, University of Waterloo	Replaces language conditioning with physical contact points. Uses VQ-BeT architecture with contact anchors. Trained on handheld gripper data, generalizes zero-shot to multiple robot embodiments.
TwinRL-VLA: Digital Twin-Driven Reinforcement Learning for Real-World Robotic Manipulation	02/2026	Online RL, VLA, Digital Twin	100% success rate across four tasks (Pick-and-Place, Insert-Hexagon-Block, Insert-Triple-Column-Block, Erase-Whiteboard). Converges in ~20 minutes with 30% speedup over prior methods (ConRFT: 77.2%, HiL-SERL: 71.25%).	Code(unofficial by Yurong Jiang)	Paper	Project	Peking University, Simplexity Robotics, Tsinghua University, HKUST	Digital twin–real-world collaborative RL framework that expands exploration space via synthetic trajectories and uses sim-to-real guided exploration to accelerate online RL for VLA models.
LingBot-VA: Causal World Modeling for Robot Control	01/2026	World Models	Real-world Success Rate (SR) / Progress Score (PS): Make Breakfast 75% SR/97% PS, Pick Screws 70% SR/82.5% PS, Fold Clothes 35% SR/48.8% PS, Unpack Delivery 65% SR/84.5% PS, Insert Tubes 40% SR/85.8% PS, Fold Pants 70% SR/76.7% PS. Achieves >20% improvement over π0.5 on challenging tasks with only 50 demos.	Code	Paper	Project	Ant Group/Alibaba	A powerful autoregressive diffusion framework (5.3B params) that predicts both video movement and actions together via Mixture-of-Transformers architecture. It efficiently plans ahead, learns quickly from data, and handles new situations well. Great at complex, long-term tasks.
Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning	01/2026	World Models	93.6% average success on challenging real-world ALOHA bimanual manipulation tasks. With model-based planning, achieves 12.5% higher task completion rate on challenging real-world tasks.	Code	Paper	Project	NVIDIA, Stanford University	Single-stage fine-tuning of pretrained video model (Cosmos-Predict2-2B) that generates robot actions, future states, and values as latent frames. Uses model-based planning with best-of-N sampling to achieve higher success rates. Can learn from policy rollout data to refine world model and value function.
Does learning from experience benefit small AI robotics models?	12/2025	Imitation Learning	4/5 when training simple ACT on imitation + corrections only.			Article		Replicating the RL loop behind Physical Intelligence's Pi*0.6 foundation model without VLAs or diffusion.
π*0.6 : a VLA That Learns From Experience	11/2025	VLA	The system ran for 13 hours straight making espresso drinks and over two hours folding novel laundry items without interruptions. Success Rates: Laundry (t-shirts & shorts) ~95%, Laundry (Diverse Hardest Items) ~70%, Make Espresso ~90%, Box Assembly ~90%		Paper	Project	Physical Intelligence	RECAP is an iterated offline RL framework that improves a Vision-Language-Action (VLA) model ($\pi^* 0.6$) by conditioning it on advantage estimates derived from a value function, allowing the model to learn and self-correct from real-world data like demonstrations, autonomous experience, and human interventions.
RL-100: Performant Robotic Manipulation with Real-World Reinforcement Learning	10/2025	Online RL	100% success across 7 tasks. 92.5% average zero-shot success on 3 tasks (without any retraining or fine-tuning), 86.7% average few-shot success on 3 tasks	Code(unofficial by Yanjie Ze)	Paper	Project	Shanghai Qizi, Shanghai Jiao Tong, HKU, UNC Chapel Hill	code to be released "after paper is accepted"
APO: Human-assisted Robotic Policy Refinement via Action Preference Optimization	10/2025	Human-in-the-loop	Improvement on success rates of Dagger, TPO, etc on in-distribution, as well as when position, background, or texture are disrupted.	Code	Paper	Project	ByteDance
HI-ORS: Human-in-the-loop Online Rejection Sampling for Robotic Manipulation	10/2025	Human-in-the-loop	Improved RW Success Rates vs vanilla BC, HIL-SERL, Q-Chunking.	Code	Paper	Project	TenCent
ARMADA/FLOAT: Autonomous Online Failure Detection and Human Shared Control Empower Scalable Real-world Deployment and Adaptation	10/2025		failure detector FLOAT achieves nearly 95% accuracy on average, surpassing prior SOTA failure detection approaches by > 20%.	Code	Paper	Project	Shanghai Jiao Tong University
SARM: Stage-Aware Reward Modeling for Long Horizon Robot Manipulation	09/2025	Rewards	83% success on folding T-shirts (flattened), 67% success on folding T-shirts (crumpled). Surpasses vanilla BC (8% and 0%).	Code (lerobot)	Paper	Project	Stanford, UC Berkeley, xdof.ai	Video-based reward modeling framework that jointly predicts high-level task stages and fine-grained progress.
Dual-Actor Fine-Tuning of VLA Models: A Talk-and-Tweak Human-in-the-Loop Approach	09/2025	VLA	100% success across three tasks within 101 minutes of online fine-tuning. For long-horizon tasks, it sustains a 50% success rate over 12 consecutive operations.		Paper	Project	Zhejiang & others	no code : (
WSRL: Efficient Online Reinforcement Learning Fine-Tuning Need Not Retain Offline Data	07/2025	Online RL	100% success rate on Franka peg insertion task in 18 minutes, SERL fails (0/20) even with 50 minutes.	Code	Paper	Project	UC Berkeley	Overall idea: No data retention during fine-tuning, warmup phase with small rollouts from pre-trained policy. Unfortunately, only 1 real world experiment, all others in sim.
Dyna Robotics (Unknown Model)	07/2025		99.9% success rate in folding towels for 8 hours/day over 3 days (dropped 1 towel on day 2). No intervention.			Project	Dyna Robotics
Figure (Helix)	06/2025		~95% accuracy at correctly orienting barcodes. 4.05 seconds per package.			Project	Figure	Adds memory for more robust, long-term tasks and force feedback for improved grip.
RSS 2025 Workshop: Human-in-the-Loop Robot Learning: Teaching, Correcting, and Adapting	06/2025		various results			Project	various universities
Compliant Residual DAgger: Improving Real-World Contact-Rich Manipulation with Human Corrections	06/2025	Human-in-the-loop	book-flipping success rate of 100% (60% improvement) and belt assembly success of 70% (50% improvement)	Code	Paper	Project	Stanford
ReWiND: Language-Guided Rewards Teach Robot Policies without New Demonstrations	05/2025	Rewards	an hour of real-world RL improves success rate from 12% to 68%, vs 8% to 10% with VLC	Code	Paper	Project	U Wash
Dyna Robotics DYNA-1 Model	04/2025		99.4% success rate in folding napkins over 24 hours. No intervention.			Project	Dyna Robotics
ConRFT: A Reinforced Fine-tuning Method for VLA Models via Consistency Policy	02/2025	VLA	96.3% avg success rate across tasks, compared to 31.9% w/ HIL-SERL	Code	Paper		Chinese Academy of Sciences	Online and offline fine-tuning.
HIL-SERL: Precise and Dexterous Robotic Manipulation via Human-in-the-Loop Reinforcement Learning	10/2024	Online RL	100% success rate on a variety of tasks	Code(official, (sim, lerobot)	Paper	Project	UC Berkeley	Online fine-tuning, human intervention allowed. Implementation available in LeRobot.
RLIF: INTERACTIVE IMITATION LEARNING AS REINFORCEMENT LEARNING	03/2024	Imitation Learning	95% success rate in cloth unfolding within 7 rounds, 100% rate success in peg insertion within 6 rounds	Code	Paper	Project	UC Berkeley
SERL: A Software Suite for Sample-Efficient Robotic Reinforcement Learning	01/2024	Online RL	100% success on PCB insertion, cable routing, object relocation	Code	Paper	Project	UC Berkeley
</task_progress>
</write_to_file>