Lambda published research on enabling 3D spatial perception for robots without dedicated depth sensors1. The work addresses limitations of monocular cameras, which most robot policies rely on and which struggle to estimate depth accurately, making precise tasks such as grasping objects in cluttered environments, inserting parts, or manipulating complex geometry difficult to execute.
Lambda Proposes 3D Robot Vision Without Depth Sensors
Lambda published research on giving robots 3D spatial perception using standard cameras, bypassing the need for dedicated depth sensors.