DARPA

now browsing by tag

 
 

Autonomous cars are Green and could be much much More

I have been reading about the progress of several manufacturers working on autonomous cars. BMW, Audi, General Motors, Volvo, Mercedes Benz, Ford, Toyota and Google are just a few of the companies looking to change the way we drive…..er move.  The claims for autonomous vehicles are simple; people behind vehicles acting in reaction to other people in vehicles can only achieve a certain amount of safety and efficiency.  Whereas vehicles acting together can increase both safety and efficiency while decreasing travel times. 

Many companies already have test vehicles roaming our streets laying the ground work for these vehicles to enter the market place in the coming decades. Although some authors and companies believe the first automated cars will be on the streets before 2020..  We shall see, in the mean time consumers are benefiting from all kinds of new automated safety and convenience features being added to vehicles every year; Anti-lock brakes, self parking cars, collision avoidance systems and smart cruise are just a few of the features that are paving the way for increased automation.  

Besides maturing the technology there are two big hurdles that hands free driving needs to overcome, first, the liability.  In the aviation industry, airplane manufacturers design to a government set of minimum standards in just about every regard.  The aircraft is then tested to show that what was designed and produced meets those minimums. The aircraft then enters service and only FAA certified professional can operate and service the aircraft.  Along the aircrafts life cycle inspections are performed to all features of the aircraft per a set of government and manufacturer guidelines.  If at any point an issue is identified the all of the aircraft can be grounded till a suitable understanding of the issue can be acquired and an approved solution found.  During the investigation of any issue liability for the repair can in most cases be easily directed back to the operator, the maintainer, the manufacturer or some combination of those parties. The recent 787 battery issue is a great example of how this system of checks and balances keeps aviation safe. In addition in the case of an issue investigators from the NTSB are able to go through maintenance log books, operator log books, aircraft black boxes and manufacturer records to identify the potential problems. They can also reconstruct an incident to determine if the source of the crash was a mechanical failure.  Lastly the NTSB tracks all reported incidents with a specific model of aircraft to identify trends. 

Getting back to the driverless car we have some issues with the liability model that cars currently operate under today.  In most cases today it is the drivers fault. The manufacturer and the mechanic are rarely targeted for issues unless clear evidence is obtainable.  But when we take the driver out of the equation whose fault does it become, the owner, the mechanic, the manufacturer?  None of these parties were at the accident so how will police assign blame, how will insurance companies assign liability?  There are no log books, there are no maintenance records, there are no certified professionals servicing your car and there is no accident investigation team out on the high way during rush hour recreating the accident. So how do we assign fault?  Without all of these checks in the system it will be hard to say whose fault it is that the car killed my family.  That first question will take a lot of thought and cherry pick the parts of aviation liability lifecycle that make sense for this more prevalent form of travel.

The second largest hurdle is transition.  If all cars are driven by people then the system operates at this level of efficiency.  If all cars are driven by machines then the system operates at an increased level of efficiency.  But what is a percentage of cars are operated by people and a percentage are operated by machines?  This is the second largest question and is what is driving a lot of my imagination.  A system made up entirely of driverless vehicles is much easier to set up and prove that it is an improvement to the current 100% people driven system.  The system that is a mix of the two takes a lot longer to set up, requires more regulation and may not have any significant impact on the efficiency of the system till a tipping point is reached.  Two states have started to look at how they would regulate such a system in hopes of luring companies and jobs to their state as this trend towards automation continues.  Nevada was the first state followed by California to get driverless car laws on the books.  Nevada, one could argue was at the heart of the driverless car revolution since it hosted the DARPA grand challenge that showed that autonomous cars could be built using today’s technology. However both states are leaping ahead in this research as automotive companies continue to set up shop in these two tech savvy states.

The laws that they have put on the books can be found here.

Nevada’s Law

California’s Law

 

So now the big question…. Which east coast state will be the first to put it’s toe into the driverless car revolution?

Crowd Sourced Engineering

Over the past 4 or 5 years I have watched the growth of crowd sourced design.  I have seen protein folding projects claim success by harnessing the crowd.  I have also seen SETI grow their search for life from harnessing thousands of computers to now also harnessing their owners. I have not personally taken part in any projects, so I went looking on the web to see any engineering design projects had used crowd sourcing and how the community perceives their success. I was able to find articles on completed software and bio engineering projects but non on any completed design projects.  However I also found a lot of engineering crowd source projects still running; NASA crowd source project, DARPA crowd source project, Open Source Rocket and a few others.  All of these projects made me a little excited and a little scared.

As a designer in the aerospace industry I am all too aware of the amount of control, checks and rechecks that go into the design and fabrication of aerospace products. The design of simple angles may be controlled by over 20 spec documents and need to be reviewed by 15 subject matter experts before it can even go out for bid to a supplier.  Once the supplier fabricates the part easily over 30 sped documents were referenced and several levels of quality check will have been performed. Lastly, the parts are continually inspected as they are received and added to the assembly or product. And even with all of those reviews 25% of a program’s budget usually goes to rework.  So this experience begs me to ask a few questions:

  • How much accountability can be built into crowd sourced engineering? 
  • How much efficiency is lost in crowd sourcing your engineering?
  • What specs, processes and controls will need to be created to make these projects viable?
  • What data formats and change management systems are used to ensure that downstream stake holders of the engineering output can plan their tasks?
  • How is functional commonality achieved? ie. how do you know all of the parts are being stress analyzed the same way?

 

I see some solutions for these questions but it will take a leader and a hierarchy to implement on the project.  Is that a viable for crowd sourced projects?  Only the future will tell I guess.

I am going to try a few of these and let you know how they turn out.