Hari Balakrishnan
Samuel Madden
Computer Science and Artificial Intelligence Laboratory
the Cartel logo
CarTel is a mobile sensor and telematics system whose design is motivated by transportation and civil infrastructure applications. Road traffic is a well-known “grand challenge” problem that affects most of us on a daily basis. For example, the Texas Transportation Institute’s urban mobility report estimates that the per-person delay caused by debilitating congestion on the highways near major cities was 54 hours on average. In addition to the loss of time and productivity, the amount of gas consumed is worth billions of dollars, and the environmental impact is enormous.
Municipalities spend millions of dollars on their roadways. Despite this investment, people are often unhappy with the quality of the roads they drive on. Poor roads are the cause of expensive lawsuits and damage claims—illustrated for example, by the more than 500,000 pothole-related claims received by insurance companies each year.
Figure 1 The CarTel system architecture. Cars on roads collect traffic and sensor data, and relay it to the CarTelHQ server via cellular or WiFi (802.11) networks. Users can connect to the server and view aggregate traffic and road quality information as well data about their own driving habits.
Balakrishnan and Madden have used a number of electrical engineering and computer science technologies, including sensing, embedded computing, and wireless networking technologies, together with machine learning and optimization algorithms, to significantly improve the state of the art in traffic planning and management. Unlike classical static sensor deployments, their approach is to use opportunistic mobile sensing—sensors deployed on cars and mobile phones). Opportunistic mobile sensing is effective because it delivers data about the roads that matter (those that are used), while being cost-effective to deploy.
The CarTel system uses in-car nodes to collect a variety of information (time, location, speed, vibration, acceleration, on-board vehicle diagnostics, etc.). This data is sent to servers using novel opportunistic wireless protocols that allow, for the first time, WiFi networks to be used from moving cars, and cellular connections (Figure 1, above). Server software analyzes the data using novel algorithms to perform tasks such as delay optimized vehicle routing (in collaboration with EECS Professor Daniela Rus and her group), carbon and emissions tracking, road surface assessment to determine which roads require immediate attention (Figure 2, below), drive data visualization, remote vehicle diagnostics, etc. Each user has an account on a commute portal (Figure 3, below) that maintains his or her data privately and provides the services mentioned above.
Figure 2 Website for the Pothole Patrol, an application that shows the location and pictures of the largest potholes in and around Boston.
CarTel is currently deployed on about 50 Boston-area taxis and a handful of MIT user cars. The team is planning a larger deployment including new algorithms and methods to services that can reduce traffic delays, fuel consumption, and carbon emissions.
The CarTel project is funded by the National Science Foundation and in part by the T-Party Project, a joint research program between MIT and Quanta Computer Inc., Taiwan, and in part by Google.
Figure 3 The drive log for one of our users. Users can see and manage the data collected about them, and visualize their driving patterns using a variety of plotting and filtering tools. (images: Balakrishnan, Madden)
Previous article: EECS Research Reaches Out – Where it makes a Difference
Next article: Autonomous Urban Driving
One Response to “The CarTel System”
-
Waco Cars says:
Very awesome! We need quantitative data and experiments like this in order to gain more knowledge on our traffic, fuel, and emissions situation. This will better our situation moving toward the future.
http://eecs-newsletter.mit.edu/articles/2008-fall/the-cartel-system/
No comments:
Post a Comment