Blog

Delivery of things

Another delivery robot company – this one operating at scale in Europe and making campus deliveries in the Valley. Starship raised $25m in June 2018. Runs on paths and sidewalks. No hints in the demo videos about how interacts with opposing direction bicycles and scooters.  Ideally something like this would be able to enter buildings and control lifts for campus deliveries. 

Commercial package delivery launch announcement (Oct 31 18) – in Milton Keynes now, and the SF Bay area before the end of the year. https://www.starship.xyz/press_releases/worlds-first-robot-package-delivery-launched-publicly-today/

Venturebeat review 

https://venturebeat.com/2018/10/31/starship-technologies-launches-commercial-package-delivery-service-using-autonomous-robots/ 

Transportation futures talk – 21 Sept 18

LondonLINXswitches

London Internet Exchange, LINX, London 1

I’m due to do a talk at the Institute of Directors in Charlotte Square, Edinburgh on the afternoon of Friday 21st September 2018, as part of the Scottish International Week. Sign up here

It will be a review of the current excitement in the Valley around #micromobility, and an exploration of parallels between data packet transport and people and thing transport.

Urban transport – Bird and Lime

limebikeJuly18

On Los Altos Avenue

Interesting to see the occasional Lime ebike in Los Altos – the nearest Lime bike location is posted as in Mountain View. No sighting yet of their Segway scooters.

For several years we tracked the increasing frequency of Teslas, and then an increasing variety of electric cars, on the streets in Los Altos. It seems like it’s almost time to do the same exercise, but for electrically assisted single person transport. Not all of them have two wheels, nor do they all have pedals. Segway sell a single wheel Ninebot. Lopfit walking bikes from Denmark have made it to Palo Alto.

These tie back into the Transportation as a System question – which company is best placed to gain scale by putting together solutions to the transportation requirements of people and things ? These personal scaled devices are part of the solution, and the travel pattern data they create have value, as indicated by the recent funding rounds. GV (was Google Ventures) led a $250m round for Lime in early June. Sequoia Capital led a $300m round for Bird (scooters) later the same month.

Updating, July 9th – the latest GV round also involves Uber, and has gone up to $335m.

Collecting background references

https://stratechery.com/2018/the-scooter-economy/ consumer surplus funded by VCs
Italian modular cars for China for services https://technode.com/2018/06/27/modular-kingdom/
Lime bikes –  https://www.limebike.com/locations
Street flow rates/capacity for cars v bikes/scooters https://www.cycling-embassy.org.uk/dictionary/capacity
Bird scooters funding explanation https://bothsidesofthetable.com/all-the-questions-you-wanted-answered-about-bird-scooters-and-their-recent-300-million-funding-62a8d557904d
Electric walking bike https://www.lopifit.nl/

Machine Learning snapshot, June 2018

Kian Katanforoosh and Andrew Ng have been teaching CS230:Deep Learning, at Stanford. The project reports and posters list has just come out, summarizing work done by the students with help from the teaching team. More than 160 projects. It will be interesting to see which of these mature into applications.

outpainting

Image from Painting Outside the Box: Image Outpainting with GANs (Mark Sabini, Gili Rusak) which was awarded first place in Outstanding Posters

CS230 https://web.stanford.edu/class/cs230/

Reports and posters http://cs230.stanford.edu/proj-spring-2018.html

The 4th Research and Applied AI Summit  https://raais.co/ just finished in London. The 125 slideset on the State of AI is a decent current snapshot of much more evolved work than the Stanford posters. https://www.stateof.ai/

Moving things

More on Transportation as a System, prompted by an article in Ars Technica.

Nuro

Autonomous vehicles for local delivery don’t need drivers; they can be smaller and lighter, slower, and at least as safe as US postal trucks.  The key to making this possible is to incent customers to collect their deliveries –  from the vehicle as it arrives, or from the nearest locker or pick-up point which can be reached by walking, perhaps with a small luggage trolley. In the US ‘rural’ post service puts letters in letter boxes which are on the street, not through your front door – but FedEx, UPS, and USPS leave packages on the doorstep.

Scheduling the delivery vehicle to arrive just when it’s convenient for the person collecting the packages to come down to the street from their multi-story building is the same kind of problem as arranging a Lyft pickup.

Solving this ‘last 50 feet’ issue of package delivery by getting the customer to fetch their packages for everything under certain sizes and weights will be much cheaper than trying to build robots to do the same job, and does not require an outside person or company to have access to the home (Amazon Key).

Amazon and Alibaba are likely to dominate global logistics; they have detailed knowledge about what their customers buy and can make supply chain predictions in order to get goods started on their journey prior to receiving specific orders. They are in a position to benefit hugely from building the overall integration of the systems for transporting goods.

Background

https://wp.cunningsystems.com/2018/03/17/transporting-people-like-ip-packets/

Nuro self driving goods vehicle https://nuro.ai/product

https://arstechnica.com/cars/2018/05/self-driving-technology-is-going-to-change-a-lot-more-than-cars/

https://www.economist.com/briefing/2018/04/26/the-global-logistics-business-is-going-to-be-transformed-by-digitisation

 

 

Edge fabric management, Facebook version

DSC06580

Omar Baldinado introducing the Facebook Networking @ Scale event, 22 May 2018

The event was held at the Computer History Museum.

Agenda https://atscaleconference.com/events/networking-scale-2018

To pick out one talk for attention, Niky Riga described the Edge Fabric which Facebook uses to configure and operate its equipment in PoPs, of which it has many hundreds, generally located in shared co-location space near to its users, so as to be able to serve content with lower latency than would be possible for direct service from the large data centers. This was a refresh of the material from an paper published in ACM Sigcomm, available here  https://research.fb.com/publications/engineering-egress-with-edge-fabric/

These events are good for meeting new people involved with networking, and for meeting up with people whom one hasn’t seen for a while, without the effort of going to Nanog or IETF meetings.

Updating, 22 June 18, to add the link to the video archive of the talks. https://code.facebook.com/posts/197834857529284/networking-scale-2018-recap/

 

Artificial Intelligence for Institutional Investors

natpressclub

Venue – the National Press Club, 13th floor ballroom

The Markets Group runs events for institutional investors. I took part in a panel about Artificial Intelligence and Machine Learning, and in a round table discussion.

Diagrams and useful background sources listed below.

Definition of Artificial Intelligence, to illustrate that machine learning is a subset of AI. Sourced from the thorough review of AI in the NHS  http://www.reform.uk/wp-content/uploads/2018/01/AI-in-Healthcare-report_.pdf

AIdefinition

Examples of AI and ML in use

At least 75% of the audience uses Netflix – which applies machine learning to improve its users’ experience of streaming as well as for content selection. Its results are driven by an extreme emphasis on keeping existing and attracting new customers. The data its customers generate by using it are used to make recommendations to them. Artwork personalization https://medium.com/netflix-techblog/artwork-personalization-c589f074ad76  Image discovery https://medium.com/netflix-techblog/ava-the-art-and-science-of-image-discovery-at-netflix-a442f163af6

View story at Medium.com

Alibaba and Tencent Unlike the US companies, they have built their support for retailing primarily based on the customer’s mobile device; so they use facial recognition for identification, image scanning and matching for item selection, precise location specification in shopping venues and integration with payment apps to enhance the customer’s buying experience, both online and in person at a store.  http://technode.com/2018/02/14/alibaba-new-retail/ http://www.bain.com/publications/articles/embracing-chinas-new-retail.aspx (joint report – Bain and Alibaba)

Recent reviews – longer reads  Recent developments are a direct result of the enormous improvement in computing capability at drastically reduced costs, a direct result of Moore’s Law – which continues.

Longer term effects of automation – 10 – 20 year horizon – Bain March 2018 http://www.bain.com/publications/articles/labor-2030-the-collision-of-demographics-automation-and-inequality.aspx

13 artificial intelligence trends reshaping industries and economies – CBInsights February 2018 https://www.cbinsights.com/reports/CB-Insights_State-of-Artificial-Intelligence-2018.pdf

Capabilities at the end of 2017 – summary in slide format from Jeff Dean (Stanford, Google Brain team) http://learningsys.org/nips17/assets/slides/dean-nips17.pdf

History

1950 Alan Turing asked “Can Machines Think”  Computing Machinery and Intelligence paper https://www.csee.umbc.edu/courses/471/papers/turing.pdf
1956 Initial definition of Artificial Intelligence at a workshop  at Dartmouth College https://en.wikipedia.org/wiki/Dartmouth_workshop   Remember the AI Winters in the 1970s, 1990s