Alibaba’s AI drives towards the smart road ahead

Profile picture for user mbanks By Martin Banks September 30, 2018
One of Alibaba's key development areas is not just autonomous, driverless vehicles but also the smart roads on which they drive. 

autonomous vehicle alibaba
Like Amazon, China’s Alibaba has found that running huge online retail operations leads to getting ever-deeper into research and development work in many associated areas. One of the key technologies here is then the development of Artificial Intelligence capabilities, for they stretch across the board, and take the companies into areas they many not at first have considered as their domain.

But there are many cross overs between domains once AI starts to take hold. A domestic robot can also be smart enough to order in fresh stocks of baked beans or gin as they become depleted. An autonomous, driverless vehicle can become part of both the supply and delivery chain of any retail organization, large or small.

Alibaba is committing heavily into th R&D of AI technology and its application, and the General Manager of its AI Open Lab, Miffy Chen, set out an overview of some of its work at the company’s recent Computing Conference in Hangzhou, China. As its name implies, much of its work is based on open source code and developments and the aim, she said, is to get everyone to pull together.

One of the key announcements she had concerned the extent to which the company has already moved into not just the driverless vehicle market but also its corollary and alter ego – the smart road. This came in the form of a fully autonomous, road-going delivery vehicle which, given the company’s predilection for prefixing product names with 'Ali’, promptly got labelled the Alivan' by your correspondent.

This is likely to prove particularly attractive in markets where there has been little previous investment in infrastructure, so it maps well onto not just China itself but also China’s increasingly close relationships across Asia and Africa. Its ability to make headway in first world countries, however, may be limited, at least in the short term.

How smart is your road?

The reason is the combination of not just a smart vehicle, which is based on Neural network-based AI and robotics technologies, and the smart road. And what is interesting about it is the market it is being pitched at comes from Alibaba’s roots firmly in the online retail business. The vehicle in question is a delivery van. In fact it received its formal licence to commence road trials from the Hangzhou regional authorities at the conference.

It is equipped with all the Lidar and Radar sensors coupled with GPS and the rest that one has come to expect with an autonomous driverless vehicle. But unlike most current car industry experiments in this area, it goes further than just being an isolated individual vehicle. It will also be able to operate as part of a `collective’ that should, in theory at least, be able to resolve issues that involve several vehicles simultaneously, including when vehicles have different priority levels.

This is where the smart road concept comes into play, and it has at least some of its roots in the high potential of Hangzhou to spawn traffic jams. The city is old – it has been big and prosperous for at least a millennium – yet like most of China it is relatively new to coping with motorised transport. Much of it was not, therefore, laid out with that in mind.

But now it is even bigger and more prosperous, not least because it is the home base of Alibaba, and huge traffic jams are the rule rather than the exception. So the city and regional authorities decided to do something about it and have invested in an extensive road management system based on a distributed network of sensing base stations located at traffic hotspots such as crossroads.

This not only senses the traffic movement around its location but can also communicate with individual vehicles. This makes it possible for the base station to manage any situation, and to effectively negotiate a pattern of movement with those vehicles to resolve problems. In this approach, the base station only needs to report a minimum amount of data back to any central management system. There is little point in continually reporting back that `every is working within parameters’.

As that reporting task is taken on by the base station, it also means that the workload expected of individual vehicle management systems can be reduced, with costs saved on individual vehicles. There will also be cost in the creation of smart roads, building out the infrastructure to support the networks of base stations, sensors and associated equipment, and this could prove to be one of the hindrances to the Alibaba approach which could work against it in first world countries.

It is still an open question as to the level of total cost and inconvenience that might be required in building out the necessary infrastructure. Most western countries already have quite sophisticated, heavily used, and generally old, road infrastructures where just the inconvenience of any build out programme could have a significant economic impact.

Third world countries, especially in areas where China is working hard to build both its influence and investment portfolio, such as the African continent, will no doubt benefit from the fact that most of the AI road and traffic management system could be built out faster than the roads themselves, and at an attractive cost in both investments and inconvenience.

It does, of course, have the potential to also be something of a two-edged sword, in that part of the traffic management system’s capabilities will be to help the state gain a better overview of the activities of its people. For example, if licence plate recognition is not part of the current scheme it is not beyond the wit of men to add it quite easily. The good side of such capabilities, of course, is the ability to adjust traffic flows so that vehicles such as ambulances get the clearest, fastest run to a hospital. There are few that would argue with that capability.

Far be it from me to suggest that it just might also be used for clearing the way for a driverless delivery van carrying same-day deliveries from any retailer.

Into the home and the factory

For some time the company has had the objective of building a new infrastructure for both business and the home based on the use of AI and robots. The company launched its first product into this area with Ali Genie back in 2016, and its first application was in smart speakers. Chen now sees a role for it in many aspects of the home.

The company used the conference to launch its latest robot technology, Tmall Genie. It communicates using Bluetooth and WiFi, and Alibaba is already working with a number of other vendors to build a wide range of domestic systems and services based on the core technology.

A number of different implementations of it are in the works – for example there is a version that is being pitched at hospitals which can dispense medicines etc to patients in the right amounts at the right times, freeing nurses for other tasks. There is also one that can carry loads of up to 30kgs.

There are also versions aimed at more domestic applications, even those required on a more grand scale. For example Alibaba is planning to open what it calls its Future Hotel which will plans to apply Tmall Genie technology to the hotel trade. This will be the first application of the technology where it will be used to control air-conditioning, passing food requests and instructions to the kitchin etc.

The idea is to provide a better, more comprehensive services to guests at all times of the day or night. This will be used as the test model for other areas of application, such as the hospital one above. And important aspect here is that it could be very useful in caring for patients with highly infectious/contagious diseases where any contact with healthcare staff carries inherent risks.

The company already has a warehouse delivery robot in operation. It is also contributing to a joint German/Chinese study on robotics in manufacturing, with a particular focus on the place of AI in combining both high volume manufacturing techniques with on-demand product personalisation by customers. One role being looked at is the use of AI as an up-selling tool, recommending add-on devices or components that can best be manufactured and included as part of the original order: a process of `you have ordered X. If you choose now, Y and/or Z can be added at the discount price of £nn’.

My take

It is, of course, useful to get some perspective on where Alibaba, and China in general, has got to in in its development of leading edge technologies such as AI, compared to the west – and it looks like it is getting rather well.

It also has good, and arguably better, contacts with the developing world where its services and technologies developments would seem to fit with the pace of emergence and development.

But it is also interesting to see their lateral thinking in action. For example, the fact that the first driverless autonomous vehicle is a delivery van, not a car. For a major retailer like Alibaba, which not only has its own logistics operation but also works closely with a host of other logistics third parties, the ability to extend the transaction data feed directly into the delivery vehicle is but a natural step to take.