New gripperThe next hardware revision, code named arpeggio, is coming in two steps. First the new gripper, which solves many of the problems identified in the previous one, most notable reducing the swing. Why isn’t the gripper just mounted at the point where the lines meet? Even though it would have the least swing, it makes the robot less useful because it wouldn’t be able to reach right next to any vertical surface such as a chair, or bed. Most commonly this comes into play when picking up laundry that fell right next to the hamper.
The second step will be the new anchor, which will switch to BLDC motors and group the spools into pairs. The two room corners with no anchor will actually have a small stainless steel eyelet through which the line passes. The new anchors have many improved features, but notably they have more strength and speed, as well as a much safer behavior during an over-torque or over-heat event.
The best part is that the new hardware revision does all this, without increasing the BOM cost.
https://youtube.com/shorts/fqnF7VVwxdM
The stringman components now all share a common raspberry pi image and require no additional setup. I just image an sd card and boot the component. This simplifies the manufacture of Stringman robots for me and saves a lot of time. In fact, there are companies that will sell you large quantities of SD cards imaged with whatever you want, so this unlocks that particular branch of the manufacturing tech tree.
Lerobot integration is coming back to Stringman - currently on the lerobot branch. With a narrowed focus on grasping. Models will learn from stabilized videos feeds and teleoperation data in a fixed reference frame so I think we will be able to see much better results from ACT and VLA models this time. This means combining the intelligence of state of the art foundation models with the reach of Stringman to make something truly capable of home cleanup at an entry level cost.
https://www.youtube.com/watch?v=-GUNNK-WNMk