Can 'smart' traffic lights ease Toronto’s road congestion?
University of Toronto team says it has developed traffic signals that 'learn' — making it possible to minimize delays at red lights.
University of Toronto Engineering graduateSamah El-Tantawy is creating better, smarter traffic lights using game theory and artificial intelligence to teach lights how to adjust to traffic patterns in real time.
"Everyone understands the impact of traffic congestion," said El-Tantawy, who was inspired by watching snarled traffic in Toronto and in her hometown of Cairo, Egypt.
"It affects the environment, the economy and society in general."
Tests of El-Tantawy's system on 60 downtown Toronto intersections at rush hour showed a reduction in delays of up to 40 per cent. The test also showed it cut travel times by as much as 26 per cent.
Existing traffic light systems use sensors embedded in the pavement leading up to the intersection to send data to and from a central management centre. The centre then sends signals back to adjust the lights' timing.
However, El-Tantawy's system processes data on site and in real time, avoiding data transmission delays. It also avoids the system-wide chaos that would result if the central management centre broke down. And while current systems monitor traffic patterns along a single road either east/west or north/south, El-Tantawy's system lets traffic lights use data from all directions. This creates more responsive timing for grid-like transportation networks. It even allows lights to 'talk' to each other to create the optimum traffic flow in a given geographical area.
"Each intersection engages in collaboration (or "game", in game theory terminology) with all the adjacent intersections in its neighbourhood where each one not only learns the local optimal control policy but also considers the policies of its neighbours and acts accordingly," said El-Tantawy. "In turn, neighbours coordinate with their further neighbours in a cascading network-wide fashion.
"In lay language, the agents act as a team of players cooperating to win a game—much like players in a soccer match, where each player endeavors to score, but at the same time considers the ultimate goal of the entire team which is winning the match."
El-Tantawy worked under the supervision of Professor Baher Abdulhai, director ofThe Toronto Intelligent Transportation Systems Centre and Testbed, to develop the system, known as MARLIN-ATSC, for Multi-agent Reinforcement Learning for Integrated Network of Adaptive Traffic Signal Controllers.
“Samah’s PhD is simply impressive, a role model for all PhD students,” said Abdulhai. “I am certainly proud of her achievements, and I hope my whole team follows suit.”
The system, which costs between $20,000 and $40,000 per intersection to install, has attracted interest from the traffic signal control industry and recently won two prestigious international awards. El-Tantawy took first place in the best PhD dissertation competition from the Institute of Electrical and Electronics Engineers' Intelligent Transportation Systems Society. El-Tantawy, who graduated from U of T in 2012 with a PhD in civil engineering, also won second place from The Institute of Operations Research and Management Sciences' George B. Dantzig Dissertation Award.
El-Tantawy has high praise for her supervisor and credits him with helping her win the awards.
"He has the critical thinking skills that made me think outside of the box," she said. "But he was not only supportive on technical matters; he also encouraged me through his positive energy."
If all goes well, El-Tantawy hopes her next step will be to help implement the technology in the field.
“I offer my heartfelt congratulations to Samah El-Tantawy for being recognized for her innovative PhD dissertation,” said Cristina Amon, dean of the Faculty of Applied Science & Engineering. “Her development of the smart traffic light control system is an excellent example of the creativity and global leadership of U of T Engineers.”
No comments:
Post a Comment