At a launch event for the school's new wireless technology research center, MIT PhD student Swarun Kumar presented technology for a new autonomous vehicle that recognizes when it may be in danger of striking other cars and pedestrians.
Several other autonomous cars have been developed elsewhere, most famously by Google, and they are generally capable of identifying objects in the road directly ahead of or behind them. The challenge undertaken by Kumar and his fellow MIT researchers is making these cars aware of dangers lurking around corners and behind buildings.
Kumar showed a video of a test run by the MIT researchers in which an autonomous golf cart running the technology, called CarSpeak, encountered a pedestrian walking from the entrance of a building to a crosswalk. The golf cart stopped roughly five yards ahead of the crosswalk and waited long enough for the pedestrian to walk to the other side of the road. The vehicle then continued driving automatically.
The solution Kumar presented is based on a method of communications that is intended to expand the vehicle's field of view. This can be accomplished by compressing and sharing the data that autonomous vehicles generate while they're in motion, which Kumar says can amount to gigabits per second.
CarSpeak interacts with the standard Robot Operating System (ROS) integrated on most autonomous vehicles to date. The ROS uses sensors to collect 3D-point cloud data that replicates physical objects in the nearby area, and a planning function to establish a path that avoids them.
Responding to the lack of ability to share this data among other vehicles, CarSpeak creates a network to access sensory information between itself and other autonomous cars and infrastructure sensors. The network could enable the cars to view CarSpeak data created over an extended area, such as moving objects in blind spots.
As noted in this report on the project, standard 802.11 networks cannot accommodate the data transmission needs for communication between autonomous vehicles because they generate more data than the available bandwidth can handle. CarSpeak instead uses a content-centric MAC protocol for transmitting data, in which data pertaining to specifically requested roads and regions contends for space in the medium, as opposed to the cars sending requests for information. This ensures the network only displays relevant data, avoiding a flood of data pertaining to open roads.
In a comparison test, a car using CarSpeak's MAC-based communications was able to stop with a maximum average delay of 0.45 seconds, compared to the minimum average delay time of 2.14 seconds for a car running 802.11, the report noted.
The key to this recognition is the way CarSpeak processes the data it receives. Because the data is so massive, effectively painting a picture of the area around the vehicle, CarSpeak places higher priority on the data that signifies nearby cars and pedestrians on the roads.
This is accomplished by organizing the information into subsets that separate the data that signifies empty roads from the data that warns of obstacles. Kumar explained it as a set of cubes that represent physical areas near the car. Those cubes are broken down into smaller cubes, or subsets, of data, which give an increasingly more targeted and accurate snapshot of the physical area.
Those cubes are assigned a numerical value of '0' if the physical space it represents is empty and a value of '1' if it is occupied by an object, like a pedestrian or another vehicle. If the cube is assigned a 0, then the entire area is empty, and therefore all the cube's subsets will be empty as well. If it is assigned a 1, then CarSpeak breaks it down to its smaller cubes and pinpoints those that are occupied by an object. The remaining empty cubes are then ignored.
This process reduces the amount of data the vehicle needs to process to just that which suggests a road hazard, and assigns that data a higher priority. CarSpeak can then process the data for occupied areas and respond accordingly without being bogged down by information that doesn't matter.
Of course, CarSpeak is not a final product, and some attendees at the event brought up remaining issues. Among those concerns were what happens when the sensors on the car become dirty - from dust or snow - or how an autonomous vehicle will respond if a pedestrian stops before fully crossing the road. Some of the human aspects of driving have yet to be addressed.
Resolving safety issues will be the objective of researchers worldwide. Kumar says that through reduced traffic congestion, higher fuel efficiency and improved productivity, autonomous cars could save $100 billion per year. The potential upside has already convinced two states - Nevada and California - to legalize autonomous vehicles, and General Motors to predict that the cars will appear on the road by 2020.
MIT has been on the forefront of autonomous vehicles research for years, and even participated in the DARPA Urban Challenge in 2007, where it competed with other universities that teamed with companies like GM, Volkswagen, and Raytheon. The opening of its new wireless center, Wireless@MIT, will only speed up the advances, as several other projects undertaken help leverage mobile technology to improve transportation safety.
Colin Neagle covers emerging technologies and the startup scene for Network World. Follow him on Twitter https://twitter.com/#!/ntwrkwrldneagle and keep up with the Microsoft, Cisco and Open Source community blogs. Colin's email address is email@example.com.
Read more about anti-malware in Network World's Anti-malware section.
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.