Risk Assessment and Fixes

Risk: Incoming sacks may be irregularly, severely tilted, or unevenly stacked, leading to visual recognition errors.

Solution: Strictly control the incoming sacks according to the following requirements.

Risk: When the sack’s edge sealing is too long, it may affect the determination of the sack’s center point, resulting in an offset of the pick point. If subsequent processes (e.g., sack opening) require precise pick points, long edge sealing may reduce picking accuracy.

Solution: Constrain the edge sealing length. If this is not possible, install a sensor before the sack opening process to locate the sack’s center point.

Risk: Incoming sacks with unsealed openings or inconsistent filling volumes may lead to visual recognition anomalies.

Solution: Enforce strict control over incoming sacks and evaluate whether a dedicated deep learning model is required.

Risk: When the sack pallet is constructed as an integrated welded tubular frame or cage, the pallet beams may obstruct the camera’s field of view during image capture, affecting visual recognition. In addition, during sack picking, the pallet uprights may collide with the robot or gripper.

Solution: Enforce strict control over incoming sacks. If such scenarios are unavoidable, conduct thorough simulation and testing in the early stages of the project, define reasonable pallet patterns and stack heights, and plan the robot’s picking path appropriately to ensure no interference occurs with the cage or frame during picking.

Risk: When the sack surfaces have striped patterns and the sacks are randomly stacked, it may be difficult to accurately segment the sack masks. This risk is relatively low for single-case sack depalletizing projects.

Solution: Enforce strict control over incoming sacks and ensure there is sufficient spacing between them.

Risk: Sacks that are not fully filled may affect palletizing stability and consistency, reduce picking accuracy during depalletizing or increase the remaining material rate after sack opening.

Solution: Remove half-filled or underfilled sacks before processing. After opening, manually handle sacks with a relatively large amount of remaining material. If the customer has no requirements for picking accuracy or remaining material rate, this issue can be disregarded.

Risk: When the surface pattern of a sack is too complex or highly reflective, it may seriously affect visual recognition. Complex patterns may cause segmentation errors, such as mistakenly merging the patterns of two tightly stacked adjacent sacks into one, or splitting a single pattern into two. Severe reflectivity on the sack surface may cause point cloud loss, ultimately resulting in inaccurate vision results.

Solution: Remove sacks with highly reflective surfaces. If sacks with complex patterns must be used, perform deep learning model training in advance and validate the model’s accuracy.

Risk: Complex and varying ambient lighting can cause sack surfaces to be overexposed or underexposed, leading to visual recognition anomalies.

Sunlight or ambient light interference in the factory may cause the sack surfaces to be overexposed.

Shadows from walls or columns may cause the sack surfaces to be underexposed.

Significant differences between daytime and nighttime lighting may result in overexposed or underexposed sack surfaces.

Solution: Provide supplemental light and shading according to the actual situation of the workstation. Please refer to Shading Solutions for details.

Risk: When a sliding device is used for the camera bracket, the low positioning accuracy of the sliding rail can result in inaccurate robot picking.

Solution: It is recommended to use a steel-frame camera mounting structure. When a sliding device is required, use a servo motor with a rail positioning accuracy of less than 0.1 mm.

Risk: When the sack stack extends beyond the camera’s field of view or working distance, it may result in incomplete image capture or poor-quality point clouds.

Solution: Mount the camera according to its working distance, or add a lifting device to adjust the camera’s height.

Risk: Using one IPC to communicate with multiple robots simultaneously leads to communication challenges.

Solution: Equip each robot with a separate IPC to ensure that one IPC communicates with only one robot.

Risk: When the requirement of vision cycle time is high, the vision system may not provide the vision result within the specified time frame.

Solution: The time from receiving commands to responding with vision information is typically less than 4 seconds. If you need to speed up the cycle time, you can mount the camera in Eye to Hand mode to incorporate the vision recognition time into the robot’s motion cycle.

Risk: High dust concentration in the sack-opening environment may affect equipment operation.

Solution: Protect the camera lens and IPC during production.

Risk: Variations in sack filling, smoothness, leaks, or wrinkles may prevent vacuum grippers from successfully picking sacks, or may cause the sacks to drop after picking.

Solution: Conduct picking stability tests on such sacks in the early stages to ensure reliable picking.

Risk: Severely tilted sack surfaces may prevent the gripper from adapting, causing pick failures.

Solution: It is recommended to use a six-axis robot for sack palletizing.

Risk: When the material inside sacks tends to clump or protrude, it may compromise the vacuum in suction cup grippers, causing sacks to drop or fail to be picked.

Solution: It is recommended to use a clamp gripper to pick sacks.