The road ahead
Updated: Apr 16, 2019
We now have a prototype that can duplicate a full haircut. And it does a surprisingly good job of it. So what next.
It's still an early prototype made of 3D printed plastic and off the shelf components. So there's still a lot of refinements and iteration to be done. The next steps are modifications to decrease the time it takes to do a haircut, and miniaturization. However we've taken it as far bootstrapping it. With the results we've achieved with the prototype, and the patent protection, we will now put together a full team and seek investment.
We think we have de-risked a lot of the concept, but there a many risks and milestones still yet to overcome.
Here is a brainstorm of the challenges we still see ahead.
Speed of haircut. The first haircut took around 50 minutes to cut. We think its possible to get a haircut down to around 6 minutes. The reasons for the long haircut time with the prototype include: We slowed the speed of it down to try to prevent damaging the prototype, which has some fragile plastic parts in. Another reason is that we could not move it forward over the head as fast as expect as the the clipper blade motor could not handle cutting as much hair at once as expected. More details on speed in the last blog.
Safety. Of course this is the biggest risk. This can only be a usable product if it can never hurt anyone. It can never pull hair or hurt anyone using it in any way. We have thought a lot about safety since its conception and have a lots of ideas for multiple layers of redundancy for safety, which we are optimistic will work.
Cost of development. This is a complex product. From the tracking system, to creating a mechanical device that is reliable and safe, this will take millions and have a long period of intense testing. So its a risky proposition. However I think if it can be pulled off, the market is immense and world wide. It's a business not just of hardware sales, but also an online marketplace for new haircut styles and hairstyle modifications.
Cutter blade life. During testing, thousands of strands of hair bunches have been cut. And we have now cut a full head of hair. So far, the clipper blades (borrowed from a household hair trimmer) seem to be holding up well. However HairSense does operate differently from a normal hair trimmer in the sense that hair is cut by the hooks pulling bunches of hair (sometimes up to 30 hairs in a bunch) up into the clipper blades to be cut. This differs from a normal trimmer which continually cuts individual hairs as it moves forward. So theres a risk that these bunches hitting the blades in groups may put an unusual force on the blade and blunt them quickly. We haven't noticed this yet but it is a risk. If this is the case, then there may be solutions to mitigate this such as harder steels, self sharpening blades etc.
Waste Hair Removal. We have done up a few prototypes of hair removal system. This removes the waste hair from the back. We ran out of time to implement any of these into the prototype so have just attached a vacuum cleaner to the device to suck the hair way. Although a vacuum is another possible solution! We don't think this is a particularly hard problem to solve, although if you are cutting your hair again after only a few weeks, the waste hair will be very small i.e. 5mm etc, so this will be a more difficult issue it will have to deal with.
Belt behavior with larger amounts of hair. During the haircut the hair removal belt worked great and successfully moved the hair from inside the device to underneath it. And for this haircut it worked fine. But we also did some tests on another mannequin to see how fast wee could move it through the hair. We moved it forward fairly fast so it would pull in around 1cm depth of hair for each revolution. The hair was captured fine, however when the belt began to turn to remove the hair, the mass of hair pulled the device forward. So this may be a problem when designing the device to pull in more hair - which it is going to need to do to move much faster through the hair. The may be solutions to this, but we have done no investigation to it. A possible solution is raising the belt higher so it is further away from the scalp, as we think the problem may be that the mass of hair is getting stuck in the very small gap between the scalp and the belt.
Tracking. The likely configuration for tracking will be a tracking bar that sits in front of the user. This bar will contain the head tracking camera and the base stations for the lighthouse system. During development, including the cutting of the full haircut, the configuration has been a RealSense camera in front the head, and base stations behind. This was done because if the base-stations where placed at the front, their lasers would interfere with the 3D tracking from the RealSense camera. From what we have read, the latest generation of the RealSense cameras are not affected by the Vive lasers.
BOM Cost. The mechanical components of the device are fairly cheap. And the price of IR cameras will be coming down considerably now they are becoming common in phones. The main cost of the SteamVR lighthouse system is the beacons, which is probably the highest cost at currently $60 US each - and there are two of these. Valve hope to commoditize these as much as possible so hopefully cost will fall quickly. It's possibly that other tracking systems may get better and could be used if cheaper. There's also the possibility that a cheaper tracking method will become available.
HairSense, in its current design, cannot cut hair short like hair clippers. I think the shortest it may be able to cut is around 10mm (which I think is a number 2 length on a normal pair of hair clippers). So currently it won't be able to give you those clean lines that a hairdresser can. This may be a deal breaker for some consumers. Possible solutions is to (1) modify the current design so the hair clippers can move forward somehow (2) add another pair of hair clippers to the device that can automatically adjust its cutting length (3) have a companion hair clipper device that is much simpler - and just a pair of hair clippers that can adjust its length as it moves over the head - this would be fairly easy to do. This would be able to use the same tracking technology and hair map as the main device.
User Interface. We think the most likely user interface to HairSense will be a mobile phone. Either plugged in or paired via bluetooth/Wifi. The phone would act as the UI to choose hairstyle, notifications etc
Hair guide. When someone is measuring a haircut or cutting hair, they need to know which areas they have already covered. The most likely solution to this will be a to have a laser shine out the front to show what areas still need to be covered. The heads location is know and the full location and orientation of the cutter, so it would be fairly easy to work out where a laser could be shone. However no prototyping work has been done on this, so theres a risk as to whether this would be usually friendly or not