Anvil Quest replaces leader-follower rigs

This is really a bet that the bottleneck in robot learning is not just model quality, it is how fast and cheaply teams can collect usable demonstrations. A Quest controller setup lets one person drive the robot directly with hardware they already know, instead of mirroring motions through a second physical arm. That matters because the whole Anvil workflow is built around long teleoperation sessions, one click recording, and fast export into training formats.

• A leader follower rig usually means buying and maintaining a second arm that copies the robot's joints. Anvil sells both options, but the Quest kit is materially cheaper than the leader follower kit, which makes it easier for labs to start collecting data without doubling arm hardware.
  1 sacra 2 anvil 4 anvil
• The practical problem with leader follower is that the operator is constrained by the geometry of the leader arm. Anvil argues that VR controllers avoid that mismatch and reduce fatigue over long sessions, while still giving direct control of grippers, recording, and replay through the Devbox workflow.
  3 anvil 4 anvil
• This puts Anvil in the same broader camp as newer teleoperation systems using Quest for data capture, but with a more packaged workflow. The product is not just robot arms, it is a bundled data factory that goes from teleop to MCAP recording to LeRobot export and model training on day one.
  4 anvil 5 sacra 6 arxiv

The direction here is toward teleoperation becoming lighter, more software defined, and easier to scale across many operators. If that shift holds, the winning robotics devkits will look less like bespoke lab rigs and more like standardized data collection appliances that can feed foundation model training continuously.