{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"The OPTIM Update","title":"Moving a Task From One Robot Body to Another | Aurora Feng, Neural Motion","html":"","width":"100%","height":180,"duration":1278,"description":"Aurora Feng is building a model that takes a task recorded on one robot and produces that same task on a robot it has never seen. Video and action together, with no retraining and nothing collected on the new body.\n \nEvery robot foundation model lab is capped by the data it physically collected. If a robot recorded 200 tasks, a policy trained on it works in the neighborhood of those 200, and the 201st is out of distribution. Aurora's argument is that there are two ways at this problem, from the data side and from the model side, and the field has spent nearly all of its attention on the model side. Co-training on every robot dataset you can find leaves the data recipe inside a black box. Her bet is that solving it at the data stage, before anything enters the pre-training pile, is what removes the ceiling.\n \nWe get into what changes and what stays invariant when the robot body changes, why kinematic retargeting solves correspondence in the wrong space, why real-to-sim-to-real loses the data on the way back, what happens to teleoperation data vendors if conversion works, and why she thinks compute is the only bottleneck left in five years.\n \nAurora is founder and CEO of Neural Motion. She founded Saturday Robotics, the largest robotics and world model research forum in Silicon Valley, scaling it from zero to 2,600 researchers in three months and recruiting most of her team out of it. Before that she was founding head of North America at LimX Dynamics, and invested at Pear VC and ZhenFund. Stanford '24.\nCHAPTERS\n \n00:00 Intro\n01:30 Building the community before building the team\n06:05 Embodiment, and what stays invariant when the body changes\n07:32 Why co-training on every robot dataset isn't enough\n10:03 Labs are dumping data across their own hardware generations\n10:58 Retargeting, humanoids, and real-to-sim-to-real\n14:25 The Q1 launch, and why China built hardware first\n16:24 Rapid fire: teleop vendors, five years out, human video\n19:33 Advice for PhDs, and what...","thumbnail_url":"https://img.transistorcdn.com/MD1uw3xsUV8NIC9WNTbAWnAxAqctNuVEgSaMzKwMDRQ/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9jYTFk/YTA0N2ZmOThmMmZl/ZGVjMjA3NWMwZDUw/YmRlMS5qcGVn.webp","thumbnail_width":300,"thumbnail_height":300}