Recently I was asked a question on Quora:

I think the external design of Mitra robot can be improvised and it can be made a little more stylish. What does Balaji Viswanathan think about it?

Yes. Of course — our designers are always working hard to make sure they can bring the best. Perfection is a process, not a state. If I got a rupee for every time people ask why don’t we design better, I would have been able to hire Jonathan Ive now. Below are the present two models we sell and we are launching even better designs in the upcoming months.

Design is hard. For instance, while at Microsoft people kept asking why we cannot design like Apple products. It is not like Microsoft could not afford designers — it is that there are a lot of engineering compromises to make.

Why is design hard?

  1. Nature of 3d: 2D design is hard — 3D design is super hard. Very few can do complex 3d objects with its complex surfaces while positioning various modules for assembly. In 2d our design looks like this below:

When we go and render in 3d, a bunch of challenges emerge and thus the result is less impressive than the 2d design.

Even the geniuses of our era had to make compromises in 3d design and it took decades to achieve the level of perfection we know them now.

2. Engineering compromises: There are a number of compromises we have to make. We would ideally want a biped design, but it is super expensive in battery life [10x more battery requirement], motor cost and reliability. We would at least want a smaller foot print, but we cannot without compromising the stability. We have to have sensors for avoiding obstacles, we have to have a “kill switch” to disable the robot if it ever goes out of control, we have to have sensors for indoor positioning and charging ports. All of this means tampering the exterior. Here you can see children playing around with the robot at the airport and trying to disturb its autonomous navigation.

Our biggest priority in design to safely avoid the children and not have any sharp part that could hurt them. That sometimes overrides aesthetics.

3. Assembly compromises: We would ideally want a unibody without any gap. However, we have to keep deploying the robots for events and sales, and these have to be done in a moment’s notice. That means we have to go for a more modular design that would enable us to unpack and pack very quickly. That modularity and the need to put panels for maintenance means more compromises in exterior.

4. Economic compromises: Almost every major manufacturer in our industry have struggled with economics. Boston Dynamics and Aldebaran had to get sold. Honda has closed its Asimo robot. Sony had huge challenges in Aibo. We are trying to avoid the fate and that requires more thought on business models and being able to change things rapidly.

When companies do in volume they use a technique called injection moulding that would produce precise plastic parts. But, the die cost alone would run into USD$1,000,000 for a robot of our size. When you change the design, you have to throw away all the die and the investments along. That is not possible for a startup like ours. Thus, we have to go for alternative moulding techniques that decrease the die cost, but at the cost of some precision.

With all these constraints in mind, we are improving the design every month. We launch new models every other month. The newer models to come next will look substantially improved from the ones above, which themselves are an improvement over what we did a year ago.