Roads in developed countries are starkly different to those in developing countries, where poor conditions cause frequent accidents, and traffic congestions increases exposure to pollution and leads to unpredictable travel times. Though these issues have been addressed by an array of Intelligent Transport Systems (ITS) that monitor traffic and road conditions and disseminate useful information, these tend to ignore the heterogeneity of vehicles characteristic of traffic in developing regions and thus fail to be of real use in those areas.
The team of international researchers that has been working on the concept of VividhaVahana, a smartphone sensor-based vehicle classification system, argues that this characteristic needs to be taken into consideration in order for ITS to reflect the specific situation in developing countries.
Shilpa Garg, Pushpendra Singh, Parameswaran Ramanathan and Rijurekha Sen arrive at four different categories of vehicles with different physical and mechanical characteristics that cause them to behave differently in similar traffic situations – two-wheeler bikes, three-wheeler auto-rickshaws, four-wheeler cars and public transport like buses. They demonstrate that their system is able to achieve above 90% vehicle-classification accuracy, and detect four vehicle classes at minimum latencies of 5 minutes, as illustrated by data collected on the roads of Delhi.
The paper further argues that VividhaVahana could benefit several common transport applications. Specifically, they argue that travel time estimates; driving pattern detection; detection of traffic states; and road surface monitoring applications could all benefit from the system, and there might be many more in the future. Though actual implementation of their proposals would require significant engineering efforts, it would undoubtedly be highly beneficial for developing regions.
The paper was presented at the MOBIQUITOUS 2014 conference that took place in London in 2014, and is now available here.
