This blog post will be describing the main challenges faced by the mech team from the beginning of pool test till now.

  1. Leakage of the main hull.

For the main hull, we have decided to use double radial o ring at the fixed end cap and sealing collar to waterproof the hull and a face seal at the removable end cap. For extra precaution, we epoxied along the edges of the fixed end cap as we would not need to remove it.


Figure 1 Main Hull

We were confident that the hull is waterproof due to the multiple layers of waterproofing and the fact that the sealing collar is being machined which has very little margin for error.

However, the water leaked into the hull through the sealing collar. As seen in the figure 1, we had no tension rods to secure the sealing collar in place. After testing the AUV for long period, heat starts to accumulate inside the main hull, causing the pressure inside the hull to increase. This pushes the sealing out such that even the inner o ring is not properly compressed against the inner surface of the hull. To prevent the sealing collar from coming out, we have design some mounts to secure the sealing collar in place.


Figure 2 Sealing Collar mounts

The top mount goes on the removable end cap while the bottom part goes to the side of the hull. We mount 2 of each on opposite sides of the hull and used cable tie to secure them. While this is able to prevent the sealing collar from popping out, it would still be better if we could have included tension rods in our initial design instead of adding last minute designs. We were lucky that there were enough space for use to mount and secure the sealing collar.

  1. Unclear front camera vision

For the front camera, it gets its vision through we fixed end cap which was self-made through gluing multiple layers of acrylic together. Not only is this end cap thick, it also contains quite a number of stains due to our lack of proficiency and experience when gluing the pieces and also due to us being a little clumsy. In the water, the camera is unable to differentiate the different colours as all the colours are very dark, resulting in it being impossible for the software team to allow the auv to detect different objects.

We then decided the best way is to create a new camera hull which requires a lot of effort.

For the camera hull, the sealing for the removable end cap is by using the screw cap. For the fixed end cap, we just epoxy the hull to the acrylic both inside and outside the hull. We also designed that the mount for the camera which is 3D printed to prevent movement of camera within the hull. It is fortunate that we heed our Captain, Quan Jie, advice to include an extra hole for spare cable gland. It came in handy as we are now able to add a separate camera housing.

The camera vision has also improved after using this clean and thin layer of acrylic.

  1. Grabber

We faced many difficulties for the grabber. The mount for the grabber on the AUV and broke off twice at different parts.

Figure 3 Grabber hinge mount

Is it not very clear in the picture but the fill for the 3D print is not 100% filled. To combat this issue, we have reprinted the mount with 100% fill.

We also faced difficulty in securing the grabber, ensuring that it is perpendicular to the ground after we have deployed it. As the AUV has the move around the bucket to try to detect the ball, the currents in the water causes the grabber to not be perpendicular to the ground all the time. The only time it will be perpendicular is when the vehicle is moving forward.


Figure 4 Grabber

I hope this picture can help you visualise. When the AUV is moving forward the current pushes against the grabber, causing the 3D printed mount to hit against the metal part circled in red. When the AUV is not moving forward, the grabber will be at a small angle away from the vertical.

We are currently testing out and fine tuning the idea of using a bolt to secure the grabber in place when it is being deployed.


Figure 5 Locking mechanism

For idea for this is that when the grabber is deployed, the bolt will drop down at the same time, and secure the grabber in place. However, in this picture the grabber is still slightly off from being perpendicular from the ground, but is at least an improvement compared to without it. We are still trying to improve on this to further ensure that the grabber will be perpendicular to the ground.

Lastly, we encountered great difficulty while trying to grab the ball. Not only, is it difficult for the software team to detect the ball, it is also quite difficult for the grabber to pick up the ball sometimes.


Figure 6 Grabber

We tried to make use of rubber bands to ‘grab’ the ball up. While, it worked quite well when we tested grabbing the ball on land, the same thing cannot be said when it is in the water. Even when we manually aimed the grabber at the ball, there are occasions that the ball did not get inside the grabber. Now imagine without our help aiming the ball, the AUV would have taken a much longer time just to pick the ball up.

To reduce the time, and also to make the lives of our software friends easier, we have decided to use a magnetic grabber instead. We have recently done the cad and got most of the parts fabricated and will be testing and assembly the parts together later today.

I wouldn’t say that we have progressed well, but at least we are heading in the correct direction.


Zhi Hao