Lv Jun (@JunLv98) profile picture

CS PhD student at SJTU

Joined June 2013
Photo by Lv Jun (@JunLv98)Photo by Lv Jun (@JunLv98)Photo by Lv Jun (@JunLv98)Photo by Lv Jun (@JunLv98)
Lv Jun retweeted
Robot Policy Iteration is bottlenecked by the "PhD-in-the-loop" requirement for data collection. We introduce 📱RoboPocket, a portable system that turns your smartphone into an intelligent co-pilot for robot post-training. No physical robot? No problem. Improve policies instantly in the wild! 🤖 🔗Project Page: robo-pocket.github.io
4
29
4
167
17,186
Imagine a future where we have a giant robot factory. How does the robot realize it has gone off track? Do we need to equip each of them with a human operator for safety's sake? Our answer is: No! Introducing ARMADA: Autonomous Online Failure Detection and Human Shared Control Empower Scalable Real-world Deployment and Adaptation! Project Website: virlus.github.io/armada/ Paper link: arxiv.org/pdf/2510.02298 Open-source code: github.com/Virlus/armada
8
55
4
311
37,011
Can robots learn from their own experiences like humans? How can they go beyond demonstrations to discover novel behaviors? ☕️Introducing SOE — Sample-Efficient Robot Policy Self-Improvement via On-Manifold Exploration! By constraining exploration to the manifold of valid actions, SOE yields diverse yet temporally coherent behaviors — enabling structured, efficient exploration. With simple rejection sampling + imitation learning, the collected rollouts fuel powerful policy self-improvement. ericjin2002.github.io/SOE
2
27
5
132
13,873
💡 With a single manually collected demonstration, can the learned policy achieve robust performance? 🤖 #RSS2025 Introducing RoboSplat: Novel Demonstration Generation with Gaussian Splatting Enables Robust One-Shot Manipulation 🔗: yangsizhe.github.io/robospla…
1
14
1
65
4,722
A good hand can push intelligence development. Introducing Eyesight Hand, equipped with full-hand high-res tactile sensors and proprioceptive actuators. It is compliant, agile, and powerful. Good tactile sensing makes learning more efficient and robust. Shout out to Branden!
7
45
3
225
23,616
Our NeurIPS'23 and ICCV'23 works on **visual reasoning**🤩: a) Symbol-LLm: LLM for the symbolic system, mvig-rhos.com/symbol_llm b) Object Concept Learning (OCL): causal relations between object category, attribute, and affordance (dataset and method), mvig-rhos.com/ocl
2
9
759
Accepted to #RSS2023!
Excited to share SAM-RL, sensing-aware model-based reinforcement learning via differentiable simulation and rendering. It allows a robot to select viewpoints and accomplish the task of needle-threading. w/Yunhai, Cheng, @shuangz, @linshaonju, @Cewu_Lu Web sites.google.com/view/sam-rl
1
8
1,310
Excited to share SAM-RL, sensing-aware model-based reinforcement learning via differentiable simulation and rendering. It allows a robot to select viewpoints and accomplish the task of needle-threading. w/Yunhai, Cheng, @shuangz, @linshaonju, @Cewu_Lu Web sites.google.com/view/sam-rl
1
2
2
10
SAM-RL updates the model via differentiable physics simulation and rendering, learns informative viewpoints and actions in simulation to accomplish manipulation tasks, and finally, applies to the real world via residual policy.
1
1
1
SAM-RL can also update the mass and inertia via differentiable physics simulation and rendering. After the action is applied both in simulation and real world, the pipeline could optimize the mass and inertia by comparing the difference between simulation and real world.
1
1
📢📢📢
Excited to share our Sagci-system #ICRA2022 w/@JunLv98,Qiaojun,Wenhai,Wenqiang,Cewu [1/2] Sagci leverages differentiable simulation and interactive perception to model environments and uses model-based learning approaches to accomplish various tasks. Web sites.google.com/view/egci
1