Archives

Date
  • 01
  • 02
  • 03
  • 04
  • 05
  • 06
  • 07
  • 08
  • 09
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31

The Allocation problem, assigning students to sessions

This is a challenge to blog readers to come up with (or find examples in practice) of good systems to allocate students to parallel sessions based on their preferences. I’ve just concluded one round of this, and the bidding system I built worked OK, but is not perfect.

The problem: Around 80 students. On 10 days over 4 weeks they will be split into 3-5 different parallel sessions on those days. Many sessions have a cap on the number of students, and more students will have them as a 1st choice than can fit. Some sessions can take many students and won’t fill up. The students can express their preference as ranking, or with numeric values.

This is known in the literature as the Allocation problem, and there are various approaches, though none I found seemed to fit just right, either being easy to code or having existing running code. But I am keen on pointers.

The goals:

  • Maximize student satisfaction/minimize disappointment. Giving a student their 1st choice is good. Giving 3rd or 4th choices is bad.
  • The system must be easy for the students to understand and use.
  • Fairness. This has many meanings, but ideally mismatches that can’t be avoided should be distributed. If somebody gets a 4th choice one day, they perhaps should have a better shot at a 1st choice on another day.
  • It’s nice if there’s a means of applying penalties to students who violate rules, sneak into sessions etc. Academic violations can result in less chance at getting your 1st choice.
  • It should be flexible. Sessions may have to be changed or many not fully finalize until a week before the session.
  • It is nice to handle quirks, like duplicated sessions a student takes only once, but where the student might have preferences for one of the instances over another. There may also be pre-requisites, so only students who take one session can have the sequel.
  • Things change and manual tweaking can be advised.

Rank sessions in order, 1st come, 1st served

This was used in the prior year. Much like a traditional sign-up sheet in some ways, students could indicate their choices in order. If more students had a session as 1st choice than would fit, the ones who filled out their form first got in. This gave priority over all 10 days and so it was changed to rotate each week to distribute who was first in line.  read more »

Robocar legality in Nevada, Germany and more news

A wrapup of robocar news from the past couple of weeks:

Nevada governor Brian Sandoval rides in Google Car

After Nevada’s recent legislation directing their DOT to explore legal operations for robocars in the state, the governor “took the wheel” of a Google car. Very positive impressions from the governor and DMV head.

Video of Virginia Tech “car for the blind”

Here’s a video interview with Dennis Hong of Virginia Tech on his effort to make a car for use by the blind.

Autonomous car team in India

A new student robocar team has sprung up in India. They’re still early but their goal of driving in the crazy Indian traffic is a daunting one. Robocars have many advantages at low speed, where the 360 degree vision of LIDARS makes them see more than a human will. Harder is modeling the behaviour of other vehicles and playing games of chicken.

More mainstream press articles

Mainstream press articles of the robocar future and the intermediate technologies are growing in number. Here’s Smartmoney on near-term technologies and a Slate piece that, like almost all mainstream press pieces, asks whether people are really willing to give up the freedom of driving. Perhaps I’m too immersed, but in my immersed perspective I have simply stopped wondering about this. There will be a few who think like the dodge ad but huge numbers of people keep asking me when they can get one.

IEEE conference at Stanford paints alternating views with optimism vs. long roadmaps

Last Saturday a small IEEE conference at Stanford covered car automation technologies, including a morning on autonomous vehicles with mixed views. Steven Shladover, for example has a decades long history in important projects like cars guided by embedded road magnets, ITS, cooperative cruise control and platooning, but he is highly skeptical of autonomous cars which drive with regular cars, insisting instead that dedicated lanes are the answer. He believes this will start by building dedicated lanes for express buses (BRT) — which is something there is political will to do in many cities — and then automating the buses in those lanes. Once this is done, cars can enter the lanes if they communicate properly with the other vehicles in the lane and the lane itself.

This infrastructure approach is simpler from a technical standpoint, but the building of new infrastructure is such a hard problem and point of slow progress that my bet, as readers know, is on robocars on ordinary streets. Without the BRT component, I view proposals for new robot-only lanes to be dead in the water. Still, it’s worth paying attention when somebody with lots of experience disagrees so fundamentally with your views.

Volkswagen, while having recently promoted their Temporary Auto Pilot, displayed a roadmap that was much slower, suggesting that having a car that could pick you up at the airport or park itself on streets was something we might see in 2028.

Another lesson from the conference was the extreme difficulty of introducing radical innovation through big automakers. Cars are perhaps the most complex product sold, as well as the most expensive consumer product for most. As a result the industry has created huge amounts of “process” to how it plans and innovates, and that process is not ready to accept much in the way of disruptive technology. As I wrote earlier about the radio as the potential place for innovation in cars, car makers are now considering the central console where the radio and other controls are found the “golden stack” and they want to be the provider of it. Especially because the stuff they sell there sells for a huge margin; people often pay $2000 for an in-car GPS that’s worse than what they get free in their phone or for $250 in the aftermarket.

German team gets permission for their robocar tests on city streets

The AutoNOMOS team at Freie Universität Berlin reports they have been approved to test on city streets. This testing will be similar to the testing Google has reported doing in California, with a safety driver and copilot in the car to monitor and take control in any situation that presents a safety risk. According to the New York Times, Google didn’t seek a specific permission but state officials did agree, when asked by the times, with the interpretation that a vehicle with a licenced driver responsible for vehicle operations was legal.

Porsche trying to make a very smart cruise control

While not up to Volswagen’s temporary auto pilot, which combines ACC with lane-following, Porsche is developing a learning automatic cruise control that will come to understand road curves and speed and drive better as it learns.

Lots of exciting news, even in the slow summer season. Disclaimer note: The Google car project is a consulting client of mine.

A solar panel on a robocar

I often see people say they would like to see solar panels on electric cars, inspired by the solar-electric cars in the challenge races, and by the idea that the solar panel will provide some recharging for the car while it is running and without need to plug it in.

It turns out this isn’t a tremendously good idea for a variety of reasons:

  • You’re probably not going to get more than a couple hundred watts of PV peak power on a car with typical cells. Even properly mounted on a roof in a sunny place like California, each peak watt delivers an average of about 5 watt-hours in a day, so 200 watts gives you 1kw-h. That’s good for around 4 to 6 miles on today’s electric cars. Not a huge range boost.
  • While thin film panels don’t weigh a lot the power they provide during actual driving would normally be only a minor boost. My math suggests they weigh more than the battery for the power they will provide while operating.
  • Panels on a car will instead be mounted flat, cutting about 30% of their output. Normally you want to tilt to the angle of the sun.
  • Cars are often in the shade, even parked indoors. Unless you work to pick your parking to have sun all day, you’ll only get a fraction of the power.
  • If you do leave your car in the sun, in many places that means it will get quite hot, you’ll burn up some of the solar energy cooling it down. (Indeed, the solar panels sometimes found on today’s hybrids and EVs don’t charge the battery, they just run a cooling fan.)
  • The worst one: If your battery is not somewhat discharged, it doesn’t have any place to put the solar energy, and so it is just thrown away. But due to range anxiety, people prefer their electric cars be kept full. It takes careful planning to use that energy.
  • A car is a very bumpy place, so you need more rugged panels than what you might put on a roof.

It is possible to get more than 200w on a car — some of the solar challenge cars that exist to be nothing but panels have gotten around a kw by using high price, high-efficiency panels. But it’s still generally much better to just put the panels on a roof where they will realize their full potential, and feed the grid, and charge from the grid.

However, on Friday I was teaching a class on the future of Robocars to my students at Singularity University and in the exercises some students wondered if they might do something for solar powered cars. (I was impressed since the students, having had only a short time to think about the issue, have to work to bring up something new.)

Robocars might solve some of the problems above, and thus possibly make sense as a place to put panels.

  • A robocar parks itself and can move. So one with a solar panel can move around to make sure it’s always in the sun, and that the sun is striking it from the right angle. It can’t move too far or too often without wasting some of the power, but it can do something.
  • When the batteries get so full that they are not making proper use of the solar energy, a robocar can find a charging station, not to charge but rather to sell excess power back to the grid and other cars. (This presumes charging stations are set up this way.)
  • Robocars could dock with other robocars that are more discharged and offer them the extra solar power, no charging stations involved — though fancy robotics are needed on the charging interface, or human beings who can do the connections.
  • If a robocar has an actuator that can tilt the panels, it can do even better. While an ordinary car could have this, an ordinary car would not have the ability to rotate in the plane of the ground to track the sun without another actuator.

It’s still not great, but it might improve things. Generally it still may be better to have the panels on rooftops and get the most from them. However, when we start thinking about super lightweight cars, cars that travel for under 100 watt-hours/mile, as well as higher efficiency panels, we might get some value if the panels are light.

It’s also expensive to install panels on top of existing facilities. Turns out that while panels are dropping below 1$/watt next year thanks to cheap Chinese capital and manufacturing, the cost of install is still over $2/watt. Cost of install on newly manufactured buildings — or cars — can be cheaper because it’s designed in from the start. The car already has the complex electrical system, while houses need to add them if they go solar.

People really are in love with the idea of a solar powered car. It’s not really possible to go green this way right now, but the future might bring something interesting.

Canada has a giant zombie in it

I discovered this year that something I’ve seen a zillion times, the standard map of Canada, features a giant, brain-eating zombie. I’m naming the zombie “Hudson” because that’s the Bay that makes up most of him. He’s a plump undead with stubby legs, a big blank eye (Price Charles Island,) and a slack jaw, and it looks like Newfoundland is in trouble.

The way the human mind likes to find figures and faces in natural patterns is called pareidolia but what surprises me is that in spite of seeing this map so often, I never saw Hudson until recently, and based on web searches, neither has anybody else.

But now I find it difficult to look at the regular map, without highlights, and not see the zombie.

He is so big he may be a cosmological zombie. They eat “branes, branes.”