New approaches for gripper design

needles above it will just hang loose after insertion. The case is worse when a foam rib runs under the gripper. In this case the needles above it will penetrate the foam, but all the other will not. If the rib is pushed down before insertion the needles above it will penetrate the foam more difficultly because of increase of compression strength.

Problem: This effect can reduce the overall needle force of the gripper considerably.

8. Needles go blunt easily.

Description: The needles used for the gripper are designed for medical use (one use only), and are sharpen that way. This means that they cut out the foam very well at the first insertion, but loose they sharpness easily. This is due to the very small cutting edge (caused by the hollow needle). The used material is surgical steel, which is semi-hard and semi-flexible. It is ideal for a needle, because they do not break easily, caused by its flexibility. The other problem is that the needles loose their sharpness in different degree that seems stochastic.

Problem: The main problem is that is difficult to determine the usability confidence of the gripper. When a needle begins to loose its sharpness, its insertion force begins to increase. The chance of it flexing also increases. It tears the foam in a larger area, making the initial hole size almost as large as the needle this way the elastic deforma-tion around the needle decreases. This causes a large decrease in its pull-out force.

9. Not enough sensors for the automation

Description: It was clear that the gripper was produced for test use only. For a real robot gripper for automating the demoulding it must be made more ‘intelligent’. Sen-sors for monitoring the working of the gripper and senSen-sors for safety reasons are es-sential.

Problem: If the gripper lacks these sensors then its usability becomes uncertain and even dangerous. For the needed sensors refer to Chapter 4.6.3.

These problems were encountered during the tests. Before trying to improve the design the problems must be analysed and possibly sold. Some problems have the same cause and a sin-gle solution is enough for many. This is why the solutions and the modifications of a new gripper design are described together in the next Chapter.

but they cannot prevent the deformation of the foam during demoulding. This way the needles cannot be placed close to the side of the mould. They must be placed at least 50-60 mms from the sides. Otherwise the deformation of the foam’s undercut (this is the part that must undergo the largest deformation during demoulding) will be prevented by the needles. In this case only the amount of foam can deform that is inside the undercut and this would result in excessive large demoulding forces. A proper needle placing will allow a portion of the free foam (the part that is not in the undercut) to deform. This will make the demoulding force considerably smaller.

One solution to take up the excessive reaction force would be to fix the gripper to the mould during demoulding to take on the counter force. This would require some kind of clamping device or powerful suction pads to fix the gripper to the mould. It would require a big and ro-bust gripper that is very complex and more difficult to operate, so we disregard this solution.

The better solution would be to push in the needles separately or in small groups. In this case the counterforce is only as large as the insertion force of on (or a couple of needles) that is much less than the overall demoulding force. This force can be taken up easily by both the robot and the gripper. This can be negligible to the demoulding force. This way increasing the needle number or making a larger gripper with more needles do not have an effect on the re-action force. Actually this was the biggest problem with the prototype gripper that caused most of the deficiencies. By making the needles penetrate separately (or in small batches) problem (3) is overcome easily. Because one needle can penetrate the foam in almost any cir-cumstances. The dent around the needle is small in this situation and reduces in size when the insertion motion stops. The foam springs back a little due to the compressed foam under the needle. If more needles are inserted together the friction force on their sides prevents the di-minishing of the dent. This way if many needles very close together are used the compression ads up, until none of the needles can penetrate the foam. Instead, they compress the foam un-der them. (Like the fakir’s pin bed.) The size of the dents around the needles are also the func-tion of the compression stiffness of the foam. The less stiff the foam is the more it compresses under the needles. If the needles are pushed in either separately or in small batches this effect would be of no importance.

The important fact is that the needles have to be further apart than the dent zone. If four nee-dles are pushed in this way, the next needle can be the one that is in the geometrical centre of the four. The friction force that acts on the side of the already inserted four needles prevents the denting of the foam under and around the newly inserted needle, so this needle can pene-trate the foam fully.

When designing the gripping the best way would be to insert the needles that are the furthest apart or situated at the perimeter of the grippe. If the gripper has the shape of a rectangle then the insertion should start with the needles on the corner. They can be inserted all together.

They not only positions the gripper and fix it to the foam but also prevent the foam from pushing in, thus allowing the inner needles to penetrate fully. The next needles to be pushed in would be those that are further from the inserted needles. These needles should be pushed in batches of four or six that have needles that are not inserted between them (See: earlier).

The next needles would be those that are situated between the already inserted needles. This should go on until all the needles are inserted. The release motion of the needles can be done the way that all the needles are extracted from the foam at the same time. This way a consid-erable amount of time can be saved. With this method the most significant problems are solved.

Making the base plate thicker can eliminate problem 4. The plate has to be sufficiently thick to guide the needles so it is also an advantage. If this plate has a thickness of at least 5mms then the needles can be fully retracted into the plate. This makes the contact of the tips of the needles and the foam brake when the needles are at the release position. If a soft foam stuck in

the guidance holes, a small burst of compressed air between the plate and the foam will re-lease the foam.

The topside of the foam has to be analysed geometrically, for grooves, ribs or any other fea-tures that disturbs the flatness (problem 7). The gripper can be design to follow the contour of the foam. Using longer/ or shorter needles where the problem is very difficult.

The foam has to be analysed carefully for inserts. It must be decided whether the inserts are at the lower part of the foam so the foam above it can be used for gripping, with shorter nee-dles, or whether they interfere completely (like inserts at the top). It is clear that a needle can-not be inserted in places where an insert is situated inside a foam (problem 6). In these places the needles must be left out from their positions. This however reduces the applicable force of the original gripper, while its weight and size remains almost the same. So instead of using a large single gripper it is better to distribute the needles in smaller gripper plates. This way it is easier to place the needles where they are needed most, so no more ineffective places will re-main. The smaller grippers can also have different shape plates to be the most effective. By letting the small grippers have individual motions will solve problem 5. The problem that only simple demoulding path can be realised by the single gripper is clear. By moving the plates separately and closely to collinearly to the axes of the needles more complex and optimal path can be realised. These paths can reduce the demoulding force considerably (Chapter 5.3.1.4) and with it reduces the used number of needles and the sufficient power and size of the robot.

This method requires complex gripper control.

If a foam part is very big (such as a car double seat) it is worth considering the use of multi-ple grippers with a large powerful robot. The use of two robots synchronised together can also be considered, however it would make the application very expensive. The needles can be kept sharp by carefully choosing a proper point. The original edge of the needle is very good but to be applied for industrial use the wall thickness of the needles have to be enlarged.

Hardened steel (instead of bainite steel) would be a better choice for the used material of the needle shaft. The hardening should not be high (around 40 HRC) otherwise the breaking is more likable. The tip should have a very hard surface to prevent the fast loss of sharpness.

One of the best methods would be to plate the tip with TiN by CVD technology. This way the sharpness would stay sufficient for the proper service time of the whole gripper.

To decrease the weight of the grippers the thick moving plate should be changed to a thinner plate with ribs. This way the same bending stiffness of it can be achieved with the fraction of its original weight.

Sensors

For the gripper to be industrial operational sensors must be added. Here only a small list of the most important of them will be mentioned.

In order to check the gripper’s status positional sensors must be attached. These can be mi-cro switches or optical sensors. These sensors tell the controller whether the gripper’s needles are fully out or retracted. Any other positions are not allowed, for both safety and operational reasons. The sensors can be attached to the pneumatic cylinder or the moving part of the grip-per (on the prototype gripgrip-per). If the needles are moving separately an optical sensor attached to the underside of the base plate can see the position of all the needles in the retracted posi-tion. A reflecting mark on the top side of the needles can signal to the optical sensor that all the needles have been inserted. If the sensor signals that all the needles are inserted only then can the demoulding take place. If the gripper is in the release stage the gripper cannot move off the release positions until all the needles are retracted and the sensor signals ‘ready’. This would also prevent the unwanted motion of the needles during robot motion. If the needles were to move during the motion the grasping would be unreliable, or if the gripper is moving without the foam the unwanted pushing out of the needles would cause a great danger.

To control the proper demoulding a six component force/torque measuring cell can be fixed between the robot and the gripper. This would provide a feedback of proper demoulding for the controller. If the demoulding-force would become too large then the motion of the robot can be reduced to avoid tearing or insufficient demoulding of the foam. In this case an error signal can be given to the operating staff. The force/torque sensor can also tell the controller when the foam has been fully demoulded, because it is known from practice, that every de-moulding is a little different. This sensor can also serve as a safety device by measuring the presence of the grasped foam’s parameters. If the foam becomes detached from the gripper (during a problematic manoeuvre) then the controller can stop the robot motion from the input signal from the sensor.

Finally a sensors should be built in that can send signals when one or more needles are bro-ken. This can only be done by a complex set of sensors that can monitor the needles in their positional matrix. The cheapest method is to fix a small inductive coil around the bottom part of each needle. The best place is the base-plate, which can be made of two sheets and the coils can be sandwiched between them. The base plate has to be at least 5 mm thick for the previ-ously mentioned reasons. If metal is between the coils instead of air the magnetic permeability increases in the coil so it shows when the needles are out. When the needles are at the top, the tips of them still reach into the coil. That is why they have to be placed further up inside the base plate of the gripper. If a needle is broken, only air will be present inside the coil. The air has a much smaller magnetic permeability, so the change can be clearly detected. By connect-ing these coils into a matrix they can show which needle in the gripper is broken so it can be found and replaced quickly and easily.