Settling Time as the Binding Constraint in High-Speed Pick and Place

The problem

In pick-and-place applications, cycle time is often assumed to be limited by the robot's maximum joint velocity or acceleration. In practice, the binding constraint is frequently settling time — the interval after the robot completes its commanded motion before the TCP is stable enough for the gripper to actuate reliably.

A typical pick-and-place cycle includes four motion segments (approach pick, retract, approach place, retract) and two precision stops (pick and place). Each precision stop requires the TCP to enter and remain within a positional tolerance band before the gripper can open or close. If the arm vibrates at the endpoint, the controller or the application logic must wait for that vibration to decay. The aggregate dwell across all precision stops per cycle directly reduces achievable throughput.

Conventional mitigations — lowering acceleration limits, adding S-curve trajectory smoothing, or extending programmed dwell timers — reduce vibration at the cost of longer motion segments or longer waits. These are effective but accept a throughput penalty rather than addressing the vibration itself.

How vibration compensation addresses this

Model-based feedforward compensation modifies the trajectory commands sent to the robot controller so that the arm's flexible modes are less excited during motion. The result is lower residual vibration at the endpoint, which allows shorter settling times without reducing the aggressiveness of the motion profile.

The approach requires characterizing the vibration modes of the specific arm in its operating configuration. This is done through a calibration procedure — controlled motions with accelerometer measurement — that produces a dynamic model of that individual robot. The model is then used to shape trajectory commands in real time.

Because the compensation is feedforward (applied to commands before they reach the servo loop), it does not require vibration sensors during production operation. It is compatible with the manufacturer's existing closed-loop controller.

Performance

• Greater than 80% vibration reduction demonstrated in testing

• Up to 2x throughput improvement by eliminating settling time overhead at precision stops

• Net cycle time improvement depends on the ratio of settling time to total motion time — applications with frequent precision stops see the largest gains

• Full-trajectory and residual-tail processing modes available depending on latency requirements

Integration

• Operates between the trajectory planner and the robot manufacturer's SDK

• No firmware changes or mechanical modifications to the robot

• Calibration uses a temporarily mounted accelerometer

• Currently supported on Standard Bots, UFACTORY, Trossen, and Denso platforms

• Containerized deployment available for production environments

Business context

Pick-and-place throughput scales directly with cycle time. In an application running 1,000 cycles per day across 260 operating days, eliminating two seconds of aggregate dwell per cycle frees approximately 144 robot-hours per arm per year. This capacity can absorb volume increases or defer the cost of adding a second arm to the cell.