The Machines that Make the Machines
The Machines that Make the Machines
This paper describes the development of intelligent robots for assembling GB300 tester trays, a complex manufacturing task requiring high precision and adaptability. The authors explore challenges such as uncertainty in part geometry and performance requirements, and present a technical strategy combining classical robotics, imitation learning, and simulation-based reinforcement learning. The work emphasizes the importance of real-world data and infrastructure for achieving reliable automation in industrial settings.
Source: Robotics — Robot learning news discovery · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
Reported success rate
95%
View original evidence
on. One Flexiv arm positions a camera for pose estimation. A second Flexiv arm inserts the limit fixture and busbar (with attached clamp). A UR arm uses an OnRobot Screwdriver to screw the busbar into the tray. The second Flexiv arm then removes the busbar clamp and the limit fixture Our busbar assembly solution achieved success rates above 95%, which we are actively pushing higher. Remaining failures come primarily
Open source S17cycle time
>124 seconds
View original evidence
ure and the busbar (with attached clamp) were initially placed at unconstrained locations on a flat surface next to the robot, and the screws initially rested within the busbar. The manufacturers requested us to achieve a 99.5% success rate and to complete the task in no more than twice the time of skilled workers (in this case, no more than 124 seconds). No unintended collisions were allowed with the tray, as even m
Open source S13cycle time
>72 seconds
View original evidence
on the board. The manufacturers again requested us to achieve a success rate of at least 99.5% and to complete the multi-connector insertion task in no more than twice the time it took skilled human workers (in this case, no more than 72 seconds for all four cables). As before, no unintended collisions were allowed with the tray. Although describing the task is straightforward, performing multi-connector insertion i
Open source S22Reported success rate
99.5%
View original evidence
ure and the busbar (with attached clamp) were initially placed at unconstrained locations on a flat surface next to the robot, and the screws initially rested within the busbar. The manufacturers requested us to achieve a 99.5% success rate and to complete the task in no more than twice the time of skilled workers (in this case, no more than 124 seconds). No unintended collisions were allowed with the tray, as even m
Open source S13Reported success rate
≥99.5%
View original evidence
on the board. The manufacturers again requested us to achieve a success rate of at least 99.5% and to complete the multi-connector insertion task in no more than twice the time it took skilled human workers (in this case, no more than 72 seconds for all four cables). As before, no unintended collisions were allowed with the tray. Although describing the task is straightforward, performing multi-connector insertion i
Open source S22number of screws
16
View original evidence
obot arms. One Flexiv Rizon 4S positioned a camera for pose estimation; a second Flexiv Rizon 4S inserted the limit fixture and busbar and later removed the busbar clamp and fixture; and a Universal Robots UR10e used an OnRobot Screwdriver for 16 screwdriving operations. In agreement with the manufacturer, the busbar clamp was lightly modified so that a single arm could unlatch it. Video 4. Our busbar assembly soluti
Open source S16number of connectors
4
View original evidence
on the board. The manufacturers again requested us to achieve a success rate of at least 99.5% and to complete the multi-connector insertion task in no more than twice the time it took skilled human workers (in this case, no more than 72 seconds for all four cables). As before, no unintended collisions were allowed with the tray. Although describing the task is straightforward, performing multi-connector insertion i
Open source S22
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ght-clearance sockets. The real-world experimental setup is shown below. Figure 1. Front view of the real-world experimental setup. Robots consist of two Flexiv Rizon 4S arms and a UR10e arm with an OnRobot screwdriver Figure 2. Overhead view of the real-world experimental setup Skilled factory workers perform these two tasks reliably and fluidly, without significant mental load. However, in another example of Morave
Open source S6
on. One Flexiv arm positions a camera for pose estimation. A second Flexiv arm inserts the limit fixture and busbar (with attached clamp). A UR arm uses an OnRobot Screwdriver to screw the busbar into the tray. The second Flexiv arm then removes the busbar clamp and the limit fixture Our busbar assembly solution achieved success rates above 95%, which we are actively pushing higher. Remaining failures come primarily
Open source S17
ure and the busbar (with attached clamp) were initially placed at unconstrained locations on a flat surface next to the robot, and the screws initially rested within the busbar. The manufacturers requested us to achieve a 99.5% success rate and to complete the task in no more than twice the time of skilled workers (in this case, no more than 124 seconds). No unintended collisions were allowed with the tray, as even m
Open source S13
on the board. The manufacturers again requested us to achieve a success rate of at least 99.5% and to complete the multi-connector insertion task in no more than twice the time it took skilled human workers (in this case, no more than 72 seconds for all four cables). As before, no unintended collisions were allowed with the tray. Although describing the task is straightforward, performing multi-connector insertion i
Open source S22
obot arms. One Flexiv Rizon 4S positioned a camera for pose estimation; a second Flexiv Rizon 4S inserted the limit fixture and busbar and later removed the busbar clamp and fixture; and a Universal Robots UR10e used an OnRobot Screwdriver for 16 screwdriving operations. In agreement with the manufacturer, the busbar clamp was lightly modified so that a single arm could unlatch it. Video 4. Our busbar assembly soluti
Open source S16
Source：Robotics — Robot learning news discovery · developer.nvidia.com