Understand the failure.
Verify the repair.
Small, reproducible software failures. The broken implementation, the fix that didn’t work, and the one that passed—preserved together.
Explore the cases ↓How results are verified ↗WHAT THE ARCHIVE CONTAINS
100840 executable cases. 20168 are open.
Every case records the implementation that fails, the fix that did not work, and the repair that passed its checks—with recorded outputs and source hashes. This release adds 100840 cases across 20168 failure mechanisms and 254 domains.
The open tier gives you the failure and the unsuccessful fix for one case in every mechanism. The remaining 80672 cases, 5 variants per mechanism, are member-only: the verified repair, its recorded checks, and the full fixture suite are held in the member archive. Read the methodology ↗
A RECORD OF WHAT WENT WRONG
Browse the archive / 100840
Self-motion term bypasses the null-space projector · case 01
Joint centering moves the end effector off the task.
Self-motion term bypasses the null-space projector · case 02
Joint centering moves the end effector off the task.
Self-motion term bypasses the null-space projector · case 03
Joint centering moves the end effector off the task.
Self-motion term bypasses the null-space projector · case 04
Joint centering moves the end effector off the task.
Self-motion term bypasses the null-space projector · case 05
Joint centering moves the end effector off the task.
Manipulability is evaluated at the absolute forearm angle · case 01
Damping switches on and off as the shoulder rotates, although conditioning depends only on the elbow.
Manipulability is evaluated at the absolute forearm angle · case 02
Damping switches on and off as the shoulder rotates, although conditioning depends only on the elbow.
Manipulability is evaluated at the absolute forearm angle · case 03
Damping switches on and off as the shoulder rotates, although conditioning depends only on the elbow.
Manipulability is evaluated at the absolute forearm angle · case 04
Damping switches on and off as the shoulder rotates, although conditioning depends only on the elbow.
Manipulability is evaluated at the absolute forearm angle · case 05
Damping switches on and off as the shoulder rotates, although conditioning depends only on the elbow.
Negative elbow angles are treated as singular · case 01
Every elbow-down pose is damped as if it sat on the singularity.
Negative elbow angles are treated as singular · case 02
Every elbow-down pose is damped as if it sat on the singularity.
Negative elbow angles are treated as singular · case 03
Every elbow-down pose is damped as if it sat on the singularity.
Negative elbow angles are treated as singular · case 04
Every elbow-down pose is damped as if it sat on the singularity.
Negative elbow angles are treated as singular · case 05
Every elbow-down pose is damped as if it sat on the singularity.
Damping ramps linearly instead of quadratically inside the zone · case 01
Joint velocities near the zone edge are over-damped.
Damping ramps linearly instead of quadratically inside the zone · case 02
Joint velocities near the zone edge are over-damped.
Damping ramps linearly instead of quadratically inside the zone · case 03
Joint velocities near the zone edge are over-damped.
Damping ramps linearly instead of quadratically inside the zone · case 04
Joint velocities near the zone edge are over-damped.
Damping ramps linearly instead of quadratically inside the zone · case 05
Joint velocities near the zone edge are over-damped.
Velocity step reports damping from the zone rather than the applied term · case 01
With zero maximum damping, steps inside the zone are labeled "damped" although they are exact.
Velocity step reports damping from the zone rather than the applied term · case 02
With zero maximum damping, steps inside the zone are labeled "damped" although they are exact.
Velocity step reports damping from the zone rather than the applied term · case 03
With zero maximum damping, steps inside the zone are labeled "damped" although they are exact.
Velocity step reports damping from the zone rather than the applied term · case 04
With zero maximum damping, steps inside the zone are labeled "damped" although they are exact.
Velocity step reports damping from the zone rather than the applied term · case 05
With zero maximum damping, steps inside the zone are labeled "damped" although they are exact.
Damped step is refused at the kinematic singularity · case 01
A straight arm returns "singular" even though damping makes the system solvable.
Damped step is refused at the kinematic singularity · case 02
A straight arm returns "singular" even though damping makes the system solvable.
Damped step is refused at the kinematic singularity · case 03
A straight arm returns "singular" even though damping makes the system solvable.
Damped step is refused at the kinematic singularity · case 04
A straight arm returns "singular" even though damping makes the system solvable.
Damped step is refused at the kinematic singularity · case 05
A straight arm returns "singular" even though damping makes the system solvable.
Damping is applied outside the singular zone · case 01
Well-conditioned poses get damped because (1-w/w0)^2 is positive for w>w0.
Damping is applied outside the singular zone · case 02
Well-conditioned poses get damped because (1-w/w0)^2 is positive for w>w0.
Damping is applied outside the singular zone · case 03
Well-conditioned poses get damped because (1-w/w0)^2 is positive for w>w0.
Damping is applied outside the singular zone · case 04
Well-conditioned poses get damped because (1-w/w0)^2 is positive for w>w0.
Damping is applied outside the singular zone · case 05
Well-conditioned poses get damped because (1-w/w0)^2 is positive for w>w0.
Trajectory dropout forgets the elbow branch · case 01
After an unreachable waypoint a branch change is no longer counted as a flip.
Trajectory dropout forgets the elbow branch · case 02
After an unreachable waypoint a branch change is no longer counted as a flip.
Trajectory dropout forgets the elbow branch · case 03
After an unreachable waypoint a branch change is no longer counted as a flip.
Trajectory dropout forgets the elbow branch · case 04
After an unreachable waypoint a branch change is no longer counted as a flip.
Trajectory dropout forgets the elbow branch · case 05
After an unreachable waypoint a branch change is no longer counted as a flip.
Trajectory branch distance ignores joint wrap-around · case 01
Near the +/-180 seam the solver picks the far branch or reports a spurious jump.
Trajectory branch distance ignores joint wrap-around · case 02
Near the +/-180 seam the solver picks the far branch or reports a spurious jump.
Trajectory branch distance ignores joint wrap-around · case 03
Near the +/-180 seam the solver picks the far branch or reports a spurious jump.
Trajectory branch distance ignores joint wrap-around · case 04
Near the +/-180 seam the solver picks the far branch or reports a spurious jump.
Trajectory branch distance ignores joint wrap-around · case 05
Near the +/-180 seam the solver picks the far branch or reports a spurious jump.
Trajectory jump test is applied to the first solution rather than the chosen one · case 01
Waypoints are rejected as jumps even though the other branch is close.
Trajectory jump test is applied to the first solution rather than the chosen one · case 02
Waypoints are rejected as jumps even though the other branch is close.
Trajectory jump test is applied to the first solution rather than the chosen one · case 03
Waypoints are rejected as jumps even though the other branch is close.
Trajectory jump test is applied to the first solution rather than the chosen one · case 04
Waypoints are rejected as jumps even though the other branch is close.
Trajectory jump test is applied to the first solution rather than the chosen one · case 05
Waypoints are rejected as jumps even though the other branch is close.
Initial branch choice is counted as a flip · case 01
Every trajectory reports one extra flip.
Initial branch choice is counted as a flip · case 02
Every trajectory reports one extra flip.
Initial branch choice is counted as a flip · case 03
Every trajectory reports one extra flip.
Initial branch choice is counted as a flip · case 04
Every trajectory reports one extra flip.
Initial branch choice is counted as a flip · case 05
Every trajectory reports one extra flip.
Trajectory output jumps by 360 at the seam · case 01
A continuous joint command steps from 179 to -179 instead of moving to 181.
Trajectory output jumps by 360 at the seam · case 02
A continuous joint command steps from 179 to -179 instead of moving to 181.
Trajectory output jumps by 360 at the seam · case 03
A continuous joint command steps from 179 to -179 instead of moving to 181.
Trajectory output jumps by 360 at the seam · case 04
A continuous joint command steps from 179 to -179 instead of moving to 181.
Trajectory output jumps by 360 at the seam · case 05
A continuous joint command steps from 179 to -179 instead of moving to 181.
Rejected jump still advances the joint state · case 01
After a rejected waypoint the next target is judged against a pose the arm never reached.
Rejected jump still advances the joint state · case 02
After a rejected waypoint the next target is judged against a pose the arm never reached.
Rejected jump still advances the joint state · case 03
After a rejected waypoint the next target is judged against a pose the arm never reached.
Rejected jump still advances the joint state · case 04
After a rejected waypoint the next target is judged against a pose the arm never reached.
Rejected jump still advances the joint state · case 05
After a rejected waypoint the next target is judged against a pose the arm never reached.
Articulated arm ignores the shoulder height above the base · case 01
Every solution is computed for a target shifted up by d1.
Articulated arm ignores the shoulder height above the base · case 02
Every solution is computed for a target shifted up by d1.
Articulated arm ignores the shoulder height above the base · case 03
Every solution is computed for a target shifted up by d1.
Articulated arm ignores the shoulder height above the base · case 04
Every solution is computed for a target shifted up by d1.
Articulated arm ignores the shoulder height above the base · case 05
Every solution is computed for a target shifted up by d1.
Back-reach solution keeps the base facing the target · case 01
The negated radius sends the arm behind a base that was never turned around.
Back-reach solution keeps the base facing the target · case 02
The negated radius sends the arm behind a base that was never turned around.
Back-reach solution keeps the base facing the target · case 03
The negated radius sends the arm behind a base that was never turned around.
Back-reach solution keeps the base facing the target · case 04
The negated radius sends the arm behind a base that was never turned around.
Back-reach solution keeps the base facing the target · case 05
The negated radius sends the arm behind a base that was never turned around.
Articulated elbow cosine drops the vertical offset · case 01
Targets above or below the shoulder get the elbow angle of a target in the shoulder plane.
Articulated elbow cosine drops the vertical offset · case 02
Targets above or below the shoulder get the elbow angle of a target in the shoulder plane.
Articulated elbow cosine drops the vertical offset · case 03
Targets above or below the shoulder get the elbow angle of a target in the shoulder plane.
Articulated elbow cosine drops the vertical offset · case 04
Targets above or below the shoulder get the elbow angle of a target in the shoulder plane.
Articulated elbow cosine drops the vertical offset · case 05
Targets above or below the shoulder get the elbow angle of a target in the shoulder plane.
Back-reach shoulder elevation uses the unsigned radius · case 01
Over-the-back solutions come out mirrored in front of the robot.
Back-reach shoulder elevation uses the unsigned radius · case 02
Over-the-back solutions come out mirrored in front of the robot.
Back-reach shoulder elevation uses the unsigned radius · case 03
Over-the-back solutions come out mirrored in front of the robot.
Back-reach shoulder elevation uses the unsigned radius · case 04
Over-the-back solutions come out mirrored in front of the robot.
Back-reach shoulder elevation uses the unsigned radius · case 05
Over-the-back solutions come out mirrored in front of the robot.
Joint wrap rounds to the nearest turn and returns +180 · case 01
A base yaw of exactly 180 is reported as +180 instead of -180.
Joint wrap rounds to the nearest turn and returns +180 · case 02
A base yaw of exactly 180 is reported as +180 instead of -180.
Joint wrap rounds to the nearest turn and returns +180 · case 03
A base yaw of exactly 180 is reported as +180 instead of -180.
Joint wrap rounds to the nearest turn and returns +180 · case 04
A base yaw of exactly 180 is reported as +180 instead of -180.
Joint wrap rounds to the nearest turn and returns +180 · case 05
A base yaw of exactly 180 is reported as +180 instead of -180.
Articulated arm rejects fully stretched targets due to rounding · case 01
Targets exactly at full extension computed through hypot report "unreachable".
Articulated arm rejects fully stretched targets due to rounding · case 02
Targets exactly at full extension computed through hypot report "unreachable".
Articulated arm rejects fully stretched targets due to rounding · case 03
Targets exactly at full extension computed through hypot report "unreachable".
Articulated arm rejects fully stretched targets due to rounding · case 04
Targets exactly at full extension computed through hypot report "unreachable".
Articulated arm rejects fully stretched targets due to rounding · case 05
Targets exactly at full extension computed through hypot report "unreachable".
Joint zero offset is added instead of removed · case 01
Every joint is commanded 2*offset away from the intended angle.
Joint zero offset is added instead of removed · case 02
Every joint is commanded 2*offset away from the intended angle.
Joint zero offset is added instead of removed · case 03
Every joint is commanded 2*offset away from the intended angle.
Joint zero offset is added instead of removed · case 04
Every joint is commanded 2*offset away from the intended angle.
Joint zero offset is added instead of removed · case 05
Every joint is commanded 2*offset away from the intended angle.
INSPECTABLE BY DESIGN
Every result has a runnable source.
Runnable implementations with recorded outputs, source hashes, and explicit contracts. Related variants share a failure mechanism and belong together in evaluation splits.
Read the methodology ↗