ARX Hand Project X1M

New design. Motors. Video

A quick impact test using a half full 2L bottle (equivalent to 1L) dropped from 1m (3.3ft) above. The test doesn't push the limits of the design by any means, but it mainly checks to see if there's any potential weak spots for general day to day use. Will test more extreme impacts at a later date when I'm able to get a better camera and mounting attachment for the hand.

All parts are printed with 2 perimeters and 10% infill except for the wrist components which are 100%.

Slow motion shot at 240fps.

Testing out the capabilities with the MK0M design in picking up objects off the desk using rubber finger cots on fingers. The design really needs a proper method of adding rubber grip to the finger tips as it would not be possible for the hand to repeat this test without.

Here's a video of how an MK0M is built. This is the first time for me editing videos so it may look jumpy or weird.

Testing the hand to see if it's capable of lifting 8kg. Fairly easy task for the hand, and I think it would be possible for it to lift even more. Will have to design something to attach to the cord as my own fingers may not fair well with a thin fishing line holding up more weight. Without the pulley system that's in the hand, there would have been no way for my fingers to bare 8kg. 

MK0M has been updated to suit either 2.85mm or 3mm diameter filament. They will allow either compatibility depending on availability of materials that can be sourced locally. Nylon is still the recommended material of choice for its durability, flexibility, low friction and low wear. Alternative materials may be used but will impact on performance and reliability.

I'm currently working together with the e-NABLE NIOP team to hopefully bring the MK0M to masses with existing e-NABLE sockets that have been developed. The time-frame for availability is uncertain but should be available in a month. So far with progress, I've made a compatible sleeve that fits their existing Quick Connect system of sockets. Transition between wrist and hand isn't particular great at the moment but may have changes in future.

Second revision helps reduce the issue that the previous prototype which had issues with strength when holding uneven loads across all 4 fingers when using a whippletree system. To help with the issue, instead of a single connection point for the main actuation cord for the whippletree block, a pulley system is used to help distribute the actuation force across the block. The pulley also has an added benefit of adding mechanical advantage to the system to help improve overall strength. 

To integrate a pulley system into the design without increasing additional part cost, the pulley system will have to function without ball bearings, and will mainly use sliding elements in the design. To use a thin fishing line that's around 0.5-0.8mm, high loads should not be focused on a small area, as it will result in wear of the surface. Without ball bearings, the cords will be sliding on the pulley surfaces, so therefore the effective diameter of the pulleys shouldn't be small as it would result in a low surface area. For pulleys that need to be small, they can be made in to pulley wheels that slide on an axle. Pulley wheels can use nylon as a low wear and low friction axle, and can be designed wide to distribute the load out onto a large surface area of the axle.

The additional mechanical advantage added to the system from the pulleys greatly increases usability of the whippletree. For how much mechanical advantage the system should have would depend on the actuation length it would result in. Having more mechanical advantage to provide greater strength is ideal, however, it adds increased length to the total length required to fully actuate the fingers. Without mechanical advantage, the fingers require around 20mm of actuation. For 2, 3 and 4 mechanical advantage, 40mm, 60mm and 80mm of actuation length is required respectively. The maximum actuation length that would be suitable for manual actuation was uncertain so a decision was made to implement a configurable 2 or 3 times mechanical advantage as it was the easiest to integrate into the system.

Initial prototype for a fully mechanical design. This prototype mainly looks into the viability and properties that a whippletree mechanism has on the actuation. 

The whippletree system that's designed into the palm consists of a rotating block with a channel for each pair of fingers, index-middle and ring-pinky.  Each channel allows for slipping between the pairs of fingers to provide independence between fingers when either is obstructed when closing. When either pair is fully obstructed, the rotating block allows for further independence between finger pairs.

The use of a whippletree allows for fingers to be independent of one another which give a more variable and conforming grip around an abject. For example in the case of holding a ball in the palm, the index and pinky will be able to surround the ball more tightly than if the fingers were dependant on one another. 

The whippletree design does however bring one downside in that it doesn't handle loads very well at extremities at pinky or index. An example where the hand may have issues is when it may be holding a bat outwards which places most of its weight closer towards the index. This issue might be due to how the whippletree can behave like a lever with reduced mechanical advantage when one side of the load is fixed. It might be possible to help reduce the effect by distributing the main actuation cord further out on the whippletree block.

Overall, the design showed promise as the whippletree was able to allow for a better conforming grip but at the cost of reduced strength against unevenly distributed loads across all 4 fingers. With the reduced strength, a pulley system for increased mechanical advantage is a must.