As AI Moves At Breakneck Speed, is Hardware Falling Behind?

Roving robots crisscross predetermined paths in Amazon’s huge warehouses, transferring parcels, reducing the need for manual labour.

Mark Hoffman
HackerNoon.com
Published in
5 min readJul 1, 2019

--

At the Re:Mars conference, Amazon revealed its new delivery rover the “Scout”, which many believe, would bring incredible efficiency to its Prime delivery model and allow the e-commerce giant to meet its promise of reducing delivery times from two days to same-day delivery.

The Scout Rover is being trained on 3D models of suburban environments that include every intricate detail the rover could come across. From the different texture of each sidewalk to even the weed plants, the rover is being made capable to move around as easily as possible.

Testing of these delivery rovers has already been started at the Snohomish County, Washington and the people behind the program say that people in that county have a favorable view of the blue and benign autonomous roving bot.

Amazon is one of the foremost firms that is leading the way towards increasing automation at all tiers of their operations. The most notable area where they’ve deployed them is their warehouses.

Small roving robots crisscross their way in Amazon’s huge warehouses, plying on predetermined paths, transferring parcels, saving manual labor.

But the Scout rover’s tires were found to wear out faster than expected. And it’s just about tires, hardware limitations may well be the biggest hurdle that developers would need to overcome in their willingness to bring in the automation led, 4th and final industrial revolution.

Amazon’s Scout is equipped with the latest AI mapping technologies that will allow it to traverse most suburban walkways and terrains, but at the end of it all, no matter how well something can see or understand, if it has no way of getting back up after falling, it will require human intervention.

With the appearance of a autonomous trolley, the Scout is limited in its functionalities. How long it can travel is limited by the power its batteries can hold and how long it takes to charge is also determined by this same inner hardware component.

The Scout can travel a mile or two at a maximum before its needs to be recharged and even if the driver takes the used ones back and then re-loads the van with Scout Rovers that have already been charged, this would mean added fuel costs for no apparently good reason at all.

The Scout Rover would need to put up more mileage than that so that the delivery van can use it in at least 3–4 neighborhoods, otherwise, if you need to go all the way back to change the Scouts to come back to almost the same spot, then the delivery rovers might not be a good fit for the Prime delivery model at all.

But there are no batteries or superchargers available these days that can pack such a big amount of an energy punch in such a small space. Tesla, the market leader in electric vehicles at the moment, has been battling with a similar roadblock since it was first introduced. Those vehicles do have charging stations over the country but who wants to waste an hour or two in the middle of the journey waiting for the car to get charged again.

This has led to slow uptake and consumer interest in these electric vehicles, making it difficult for governments and manufacturers to slow down the use of fossil fuel based vehicles and replace them with cleaner, electric ones.

Coming back to the Scout, the problem isn’t only with the battery, but also with how it cannot behave like a human unless it has hands. The main idea behind bringing in automation is to remove menial human labor from the equation thus powering efficiency.

But if the Scout cannot pull out the parcels out of its own storage and chuck it through the mail box or even leave it at the doorstep of the customer, it would need to wait till either the customer gets there physically to remove the parcel or it could leave and come back later.

Both of these scenarios are wasteful and make all that AI led mapping technology which allows the Rover to roam around safely, go to waste.

And currently there is little chance that the developers can add an anthropomorphic arm on to the Scout and that’s because, to do that you would need to use a slip ring and those currently come in such big sizes that they can be a fit for an industrial sized automated machine but not for this portable, small delivery rover.

If you are thinking that why doesn’t the Rover just open its storage box from underneath and just drop the parcel? That’s because doing this can damage the inner components of the parcel, which could contain anything from a mobile phone to a vase. So that idea is certainly out of the scenario and we definitely need an arm if the rover has to be able to do that.

On top of this, what if the Rover falls over? A very real scenario for anything that moves, there is no way a Rover could re-correct itself and continue on its way without human intervention.

Amazon is currently sending people with the Scout known as Ambassadors, but shouldn’t moving a rover around be different from walking a dog?

If we want to build a society on automation, the over-reliance on developing the software part of it should be supplemented with equal attention on developing the hardware components involved.

Till then, the astonishing advancements in AI will only be effective in areas where the handicap from hardware parts doesn’t play a big role in the effectiveness that can be achieved, like in detecting cancers or fractures more efficiently or in bringing more authentic rankings on search engines. But as for the challenges of the real physical world, we might be well ahead in our journey towards developing a better brain — but we might have to wait a while to find suitable bodies to put them into.

--

--