Robotic Gripper for a Kiwifruit Packhouse
A Bay of Plenty packhouse is trialling a robotic arm to pick and place kiwifruit from a conveyor into trays. The arm must handle 120 fruit per minute, with a bruise threshold of 2.5 newtons of contact force, and it operates next to a manual packing line where workers move between stations. The current gripper is a two-finger pneumatic design with soft silicone pads and a force sensor. Trial data over two days shows a throughput of 96 fruit per minute, a bruise rate of 4.2% at the stem end, and three near-miss events where the arm continued moving after a worker entered the cell. The safety scanners are rated IP65 but the packhouse washdown creates standing water and steam, and the scanners logged four false triggers per shift. The packhouse manager wants to decide whether to continue the trial, modify the gripper, or change the cell layout. The budget allows one more month of engineering time before the fruit season peaks.
1. Identify the most serious failure mode in the trial data and justify why it should take priority over throughput and bruising.
Read the worked answer
The most serious failure mode is the three near-miss events where the arm continued moving after a worker entered the cell. Throughput at 96 fruit per minute is below the 120 target, and the 4.2% bruise rate is a quality problem, but both are economic losses. A near-miss is a safety failure that could cause serious injury, and it indicates that the safety system is not reliably detecting people. The false triggers on the scanners are not the same as a failure to stop: four false triggers per shift suggest the scanners are working but are being disrupted by water and steam, while the near-misses suggest either a detection gap, a software inhibit, or a cell layout problem where a worker can enter a blind spot. The priority is to establish why the arm continued moving. If the scanners are IP65 but the connectors are not, water ingress could cause intermittent loss of signal. If the cell layout allows entry from the side, the scanner field may not cover that path. Until this is resolved, the trial should not continue with workers present. The manager should treat the safety issue as a stop-work condition, not a trade-off against throughput. Bruising can be reduced by softer pads or lower force, and throughput can be increased by tuning the motion profile, but neither matters if the cell is unsafe.
2. Recommend a technical change to the gripper or cell that would address the bruising without reducing throughput, and explain the trade-offs.
Read the worked answer
The bruising is concentrated at the stem end, which suggests the two-finger gripper is applying a point load near the stem when the fruit is picked from the conveyor. A technical change is to replace the two-finger gripper with a three-finger or a soft vacuum gripper that contacts the fruit on the sides rather than the stem. A three-finger design distributes force over a larger area and can centre the fruit, which should reduce the contact pressure below the 2.5 newton threshold. The trade-off is that a three-finger gripper may be slower to open and close, which could reduce throughput, and it may require a different approach path. A vacuum gripper could be faster but may not work well on the fuzzy kiwifruit surface and may leave marks. Another option is to keep the two-finger gripper but add a compliant wrist and a force-limiting valve, and to change the pick height so the fingers close on the widest part of the fruit. This is cheaper and does not change the cycle time much, but it may not eliminate stem-end bruising. I would recommend prototyping the three-finger gripper with soft bellows pads and testing it on 500 fruit at the current cycle time. If the bruise rate falls below 2%, the throughput can be recovered by optimising the conveyor tracking. The trade-off is engineering time: one month is tight, so the team should run the gripper trial in parallel with the safety fix.
3. Explain how you would redesign the cell layout and safety system to prevent further near-misses, and what evidence would justify restarting the trial.
Read the worked answer
The cell layout should be changed so that the robot and the manual line are physically separated by a fixed guard, with a single entry gate controlled by an interlock. The current layout allows workers to move between stations near the robot, which creates unpredictable entry paths. A fixed guard removes the blind-spot problem, and the interlock ensures the arm stops when the gate opens. The scanners should be retained as a secondary layer, but they should be mounted away from washdown spray and fitted with air curtains or heated enclosures to prevent condensation. The false triggers suggest the scanners are seeing water droplets, so a change from IP65 to IP69K-rated scanners may be needed, or the scanners can be replaced with light curtains that are less sensitive to steam. The safety logic should be validated with a fault-injection test: simulate a scanner failure, a gate opening, and a worker entering from the side, and record the stop time. The evidence to restart the trial should include a documented risk assessment, a stop-time measurement under the worst-case load, and a trial with no worker entry for at least 200 cycles. The packhouse should also involve the workers in the layout review, because they know where people actually walk. Restarting without these changes would be unsafe, and the near-misses show the current system cannot be trusted just because the scanners log false triggers.