A global archive of independent reviews of everything happening from the beginning of the millennium

Read our Copyright Notice click here

For publication dates click here

Driverless Cars 2013

Driverless Cars 2014 (January-June)

Driverless Cars 2014 (July-December)

Driverless Cars 2015 (January-June)

Lane Markings

Cambridge Autonomous Mobility



4 August 2015

Driverless cars will need new types of mapping and if an intolerable burden is not to be placed on highway authorities or satellite services to continually update maps some of the mapping will have to be done by the eventually growing fleet of autonomous vehicles in real time using 'swarm intelligence'. Cloud services will also have to provide a backbone of up to date maps which the navigation systems of autonomous vehicles can reference.

Nokia HERE already provide data to a large number of digital map suppliers like Garmin and Bing.

They are also developing maps that are only readable by computers, not humans, and these will be used by driverless cars. These are being produced with the aid of vehicles fitted with sensors that acqiure data for the maps.

Yesterday, Audi, BMW and Daimler jointly announced that they intend to acquire the HERE mapping and location services from Nokia and to complete the acquisition in early 2016, at a cost of
2.8 billion, and to make its maps available to all automobile makers, suppliers and customers on an open basis.

The sub-text, one might presume, is that automotive maps will not standardize around Google maps.

Computer generated HERE image with sensor assisted compilation

A BMW press release states:

High-precision maps are important for autonomous driving and many other forms of assistance systems, as these technologies require an up-to-date plan of a vehicle’s surroundings exact to the nearest centimeter, in order to react in real time. While HERE already produces extremely precise static maps, they can be verified more exactly and continually updated with a constant flow of data from vehicles’ surroundings;

and this on swarm intelligence:

The social benefits of swarm intelligence are enormous: They facilitate warnings of hazards in real time, of icy roads for example, based on calculations of individual data such as ABS activations and outside temperature. Upcoming traffic jams will be identified more precisely in the future, significantly reducing the risk of accidents. In this way, the vision of accident-free driving is gradually becoming reality. In a further stage, the data could be used to learn about critical bends on the road, in order to warn drivers in good time or to activate assistance systems. Anticipation of green phases of stoplights could navigate vehicles through an urban area on a “green wave” with the appropriate engine performance and minimized fuel consumption.


Further HERE image issued at completion of acquisition - 14 December 2015


16 September 2015

One of the things about the consumer success of smartphones and tablets since the first iPhones is that we have grown accustomed to controlling everything from screens and the more traditional buttons.

The internet of things, for all the hype surrounding it, is likely change things.

We open the curtains by fractional gradations by pulling on a cord. I would like to see it being possible to pull on a cord of one curtain of a south facing window to let a little light in and have all south facing window curtains do the same. Of course, we could do this with an ever-so-serious app on a tablet but the physical delight of pulling the cord and seeing how much light is coming in and where it is falling is not something that needs to be renounced.

The driver-vehicle interface and how important it will be for driverless cars has been touched on here before. Google came in for some subtle ribbing for equipping its first proprietary prototype driverless car with a single button.

Now it has changed this and has told Forbes:

It's the first vehicle that's designed just for self-driving. It has no steering wheel, no brake pedals, no manual controls at all. All it has is a 'go' button, a 'please slow down and stop' button, and 'stop really quickly' button. The intention is for the passenger to get into the vehicle , say into a microphone take me to ........., and the car does the entire journey for you.

The early part of this statement is fine. Any vehicle manufacturer which thinks a true self-driving car has to have a steering wheel and brake pedal is mistaken. You do not sit in the back of the bus with a steering wheel and brake pedal 'just in case' the driver has a heart attack. There will be self-driving cars that offer these for reasons of driver satisfaction but that is another matter.

The new parameters indicate a way forward. There is going to be some element of voice control by the person nearest to the microphone (assumming his or her voiceprint has been recognised - you will not be wanting infants or the parrot giving commands to the car) and the three button option gives more reassurance than a single 'go' button. Yet this is not anywhere near the sophisticated driver-vehicle interfaces that traditional car manufacturers know something about.

Screens are a mainstream option but can be slow to navigate especially if you are on another page which is why professional photographers prefer cameras with lots of buttons and dials but this presupposes you know what the buttons and dials are for and where they are.

So how do you get passengers to control the car in transit when required? How do you tell it 'go round the block, I'm looking for the dog'? Or how do you sound the horn when a child is weaving in front of the car on a bicycle? You need to make a decision on whether to sound it dependent on context, not leave it to the car. How do you simply say ' turn left now' - which is not in the Google three button command set?

Arguments have been advanced against full steering wheels and joysticks. Are the ones against minature steering wheels and joysticks for emergency purposes yet conclusive?

So perhaps passengers could have something like a marshal's baton or wand in each of the door pockets. By its orientation, or with a rotatable end like on a motorcycle, you could turn the vehicle left or right, press a button for an emergency stop, another for a U-turn, another to open the tailgate, another to turn it on as an emergency light and so on.


25 October 2015

I am of the view that data traffic required for autonomous vehicle navigation should be carried on separate radio frequencies to those used for other forms of communications to and from the vehicles e.g. for infotainment, voice, text and email.

The 5.9 GHz DSRC frequencies should be used and all the data traffic should be encrypted.

In an ideal world, but stated here only for the sake of the argument, all internet of things data exchanges within buildings should also be on new frequencies and not use existing wifi frequencies (and not the DSRC frequencies reserved for automotive use).

There are a number of reasons for this and the list given below is not exhaustive.

1. Hacking of the automotive frequencies becomes much harder.

2. The providers of autonomous vehicle services will implicitly have knowledge of the position of vehicles, where they stop and so on, as part of their guidance assistance but this should not mean that they can at the same time in some manner read the personal traffic of the vehicle occupants. So the provider might know that a driverless car is stopping at a traffic light but not know that an email was being sent at that traffic light by the passenger in the rear left seat simply by virtue of the nature of the traffic being sent over the navigation frequencies.

3. The same holds true with the internet of things. You do not, for instance, want the electronically commanded closing of a curtain in a particular room at a fairly regular time to imply to a third party that a child sleeps there. Separate frequencies reduce the risk.

5. There would be much less noise from other data packets on the navigation frequencies with obvious benefits in detecting intrusions.

6. As the recent hack of TalkTalk servers in the U.K. shows we have made a mess of both security and privacy on existing communications channels like the internet, wifi and mobile. Only traditional landline telephony remains relatively unpolluted. We should start afresh with new services like the provision of autonomous vehicles.

7. In using old technology we may rightly object to changes that bring intrusions not bargained for. Someone may have bought a motorbike for the freedom, or at least the illusion of freedom, that it gives. Legacy motorbikes, indeed all legacy vehicles, I would argue, should not be obliged to carry DSRC beacons for a small gain in safety.

On the other hand, where we espouse a new technological package, like driverless cars, we should accept its implications. One is probably that at least one entity will know where the vehicle is.


20 November 2015

The question is what data do we want to go over 5.9 GHz DSRC (potentially very secure, potentially all encrypted and with very low latency); 4G mobile networks (quite secure and low latency); WiFi routers in buildings and in cars (not so secure and often with latency problems) and over the internet in transit (awash with third party tracking).

So you might want to send intelligent transport system (ITS) data and autonomous vehicle guidance data over 5.9 GHz DSRC; mobile voice telephony, eCall and locally provided updates of navigation maps over 4G; email, infotainment and web surfing over WiFi and the internet.

Also taking the widest overview of all, that architects are trained to have, what type of communications within buildings, and between buildings and vehicles in near proximity to them, should there be to enhance privacy in domains that have no reason to have their data sent over the internet or be subject to third party tracking?

My preference is LiFi but we do not have that technology yet. WiFi is the porous first generation alternative we are using now.


21 November 2015

Integrating technology in service, BMW showcased premieres of some new apps at IFA 2015 (Berlin 4-9 September 2015) on the i3 electric vehicle.

Deutsche Telecom's Smart Home apps allow users to control in-building functions, such as heating, from a car. Users can check whether windows and doors are locked or the burglar alarm is activated while on the move. Users get notified on their car's dashboard about water leaks or intruder alerts via the integration.

Samsung's SmartThings research app allows functions in connected houses, such as automatically ensuring your doors are locked and temperature is adjusted as you drive away from the house. The depth of BMW ConnectDrive app and the ease of integrating it into the SmartThings platform deliver the connected car experience to both Android and iOS users.

(Source: BMW)


19 December 2015

The state of California has published new draft regulations for autonomous vehicles that take onboard many of the issues raised here over the past year.


By the end of 2015 a number of things were apparent:

Pods running offroad would not have software permitting them to operate autonomously;

the legal classification of autonomous vehicles in the same legislative space as internet of things devices would be a mistake;

any attempt to classify autonomous vehicles as drones displayed a fundamental lack of understanding of the challenges civilian autonomous vehicles would face if they could be made a reality.

It was time to become more hard-nosed.

Lane Markings