Data61 helping NSW predict traffic incidents using artificial intelligence Posted on : Feb 15 - 2019

The Commonwealth Scientific and Industrial Research Organisation's (CSIRO) Data61 has partnered with Transport for New South Wales (TfNSW) to help improve the efficiency and effectiveness of transport systems.

The Traffic Congestion Management program, which has been under way for a few years, is touted as "analysing automated end-to-end, multi-modal journey planning for operators and passengers".

Addressing the Standing Committee on Infrastructure, Transport and Cities' inquiry into automated mass transit in Australia on Friday, Dr Chen Cai, leader of the Advanced Data Analytics in Transport (ADAIT) group at Data61 and the manager of Data61's intelligent transportation system (ITS) business, said the partnership has seen the development of a prototype artificial intelligence (AI) engine for congestion management.

According to Cai, by using the tool, it is possible to model the impact of network changes or disruptions and then issue automatic journey planning information for transport operators and travellers.

"[It's about] how to bring intelligence into managing next-generation congestion, with the key idea, instead of being reactive, which is how we used to manage events in road networks, we want to be proactive," he explained.

"We want to predict what's going to happen in the future from now."

The work is centred on determining what is going to happen on the road, through using the information both organisations have accrued up until now, including real-time, historical, and passenger transport data.

"We came up with a self-learning engine, an AI engine, that can learn from what happened prior to this date," Cai said, noting that this includes insight into thousands of accidents on NSW roads.

"We have a fairly good understanding of what are the reoccurring events, so if similar happens, we know what is the best way to have -- that's where the AI engine would recommend to the operator, and once the engine detects the scenario, these are the probably outcomes and these are the most appropriate actions to handle the situation."

Information trawled through by the AI engine includes data generated by tens of thousands of detectors throughout the road network, coming from Opal, GPS devices, traffic signals, and buses. It also includes emerging sources such as mobile phones, which Cai said provides vehicle speed and congestion-related insight. View More