Stephan Winter winterunimelb edu au Smart Transportation needs

Stephan Winter winter@unimelb. edu. au Smart Transportation needs Computational Transportation Science

Computational Transportation Science © HERE • an emerging discipline that combines computer science and engineering with the modeling, planning, and economic aspects of transport. • studies how to improve the safety, mobility, and sustainability of the transport system by taking advantage of IT and ubiquitous computing. Wikipedia (accessed 11/2015)

Computational transportation science © HERE • coll. : “The science behind intelligent transportation systems” Winter, S. ; Sester, M. ; Wolfson, O. ; Geers, G. (2011): Towards a Computational Transportation Science. Journal of Spatial Information Science, 1 (2): 119 -126. • but then: what are “intelligent transportation systems”? Turing, A. M. (1950): Computing Machinery and Intelligence. Mind, 59 (236): 433 -460.

Intelligent transportation systems © HERE • “advanced applications which, without embodying intelligence as such, aim to provide innovative services relating to different modes of transport and traffic management and enable various users to be better informed and make safer, more coordinated, and 'smarter' use of transport networks. ” Wikipedia (accessed 11/2015)

Intelligent transportation systems But then: © HERE Vehicles and people interact • “advanced applications which, without embodying intelligence as such, aim to provide innovative services relating to autonomous vehicles different modes of transport and traffic management and connected vehicles enable various users to be better informed and make safer, more coordinated, and 'smarter' use of transport sharing, platooning, networks. ” progressive signal systems, Wikipedia (accessed 11/2015) dynamic pricing, mode integration, etc.

Computational Transportation Science Computer Science • • Travel demand modeling Traffic control Transportation safety Traffic flow and capacity Automated vehicle control Routing and network models Scheduling and optimization Theoretical computer science o • … Applied computer science o o o o o Artificial intelligence Computer architecture and engineering Computer performance analysis Computer graphics / visualization Computer security and cryptography Computational science Computer networks Concurrent, parallel and distributed systems Databases Software engineering

Computational transportation science © HERE • Core elements – goal-directed, time-constrained movements of independent agents • people, goods – instrumented by vehicles of various modes and physical constraints – bound to transport infrastructure of other constraints • guideways, terminals, control policies • Derived elements – flocks, crowds, queues, platoons – events and processes – flow / capacity agents & vehicles are sensor-rich and connected

• embedded sensor platforms – vehicles, travellers, infrastructure • black box, smartphone, CCTV, SCATS, e-toll, smartcards, … – grounding transport simulations Oak Ridge National Lab. HERE Computational transportation science – distributed, mobile systems of unprecedented scale • streams (“big” data), integration (semantics) analytics (data mining) – local or central coordination, collaboration MATSIM / Senozon • connectivity – informed decisions – cognitive engineering Ronald et al. 2015 • interaction

Computational transportation science © HERE • After all: a science? – science: yes – a science: no. • At the intersection of disciplines – transport/mobility: complex • travel: derived, not for its own sake – domain of computational challenges • data (availability, accuracy, timeliness, suitedness) • correlation with structure / design / behaviour of world • use (economic, environmental, and social impact)

Examples from Melbourne Ad-hoc demand-responsive transport • shared • point-to-point • ad-hoc 1. 2. 3. 4. 5. Some results Exploring susceptibility of shared mobility Ridesharing with social contacts Modelling mobility by DRT vs PT Exploring on-demand co-modality Let’s share!

Some results • Simulation platforms need Nicole Ronald extensions for ad-hoc DRT • Scenario testing, Richard Kelly e. g. , feeder services Michael Rigby • User interaction: launch pads for service areas

Exploring susceptibility of shared mobility Shubham Jain Objectives – find an estimation of demand patterns for DRT based on demography – avoid service-specific survey Methods – review of usage patterns of some of the existing DRT services in different regions of the world – analysis of socio-economic demography – analysis of current trip characteristics from household travel surveys (VISTA) – predict usage patterns Susceptibility of shared mobility in Greater Melbourne

Ridesharing with social contacts Yaoli Wang Objective – Can uptake be increased by matching with friends Why – High overlapping rate of individual trajectories • 40% trip length saved in NYC 1 • 70% ~ 88% of surveyed MIT students and staff can share rides 2 – Socio-psychological barriers 3, 4 • 7% detour tolerance for strangers vs. 30% for friends • 10% willingness to ridesharing with strangers vs. almost 100% for friends 1. 2. Santi, P. , Resta, G. , Szell, M. , Sobolevsky, S. , Strogatz, S. H. , Ratti, C. , 2014. Proc. Natl. Acad. 3. Sci. 111(37), 13290– 13294. Amey, A. M. , 2010. Thesis: Massachusetts Institute of Technology. 4. Key findings – Ridesharing only with social contacts does not necessarily increase detour cost – An algorithm giving priority to friends significantly increases the matching rate between friends • 14% ~ 15% of the total population (increased from 3%) • average uptake rate in simulation is 38% (increase from 24%) Chaube, V. , Kavanaugh, A. L. , Pérez-Quiñones, M. A. , 2010 In: Proceedings of the 43 rd Hawaii International Conference on System Sciences (HICSS). Honolulu, HI. Wessels, R. , 2009. Combining ridesharing & social networks.

Modelling mobility by DRT vs PT Zahra Navidikashani Objective: Could DRT be the solution for unprofitable conventional public transport (PT) in low-demand areas? Method: Using an ad-hoc dynamic routing algorithm embedded in the Multi Agent Transport Simulation (MATSim) software package Tested: Replacing PT with DRT Reduction in perceived travel time Increase in mobility DRT cost Grid network provides a better environment for DRT operation.

Exploring on-demand co-modality Nicole Ronald Demand-responsive transportation and delivery of on-demand food Co-modality 1: same scheme, different vehicles Co-modality 2: shared vehicles Bus and van icons made by Freepik from www. flaticon. com (CC BY 3. 0) • Findings: – Combining schemes leads to improved performance for both passengers and deliveries – more resilient to uneven or unexpected demands • Currently developing optimisation methods for improved performance

Discussion – Conclusions SMART TRANSPORTATION NEEDS CTS

Unintended consequences of CTS • information overflow → interference – stress – errors • denying choice → surrender – – • sedentary locomotion serendipity engagement learning responsibility → health – obesity – diabetes Norbert Wiener (1948): Cybernetics

Conclusions © HERE • ‘smart’, ‘intelligent’ are vague terms here • transport is a complex systems (within the complex system city) • computational aspects can never be one-dimensional • evaluations must consider the full system impact

© Copyright The University of Melbourne 2015

Sensors: • vehicles • travelers • infrastructure Connected devices: • Transportation systems, due to their distributed/mobile nature, can become the ultimate testbed for this ubiquitous (i. e. , embedded, highly-distributed, and sensor-laden) computing environment of unprecedented scale. • if they are to be made available in real-time to wireless devices such as cell phones and PDAs How people will engage: • numerous novel applications • order of magnitude improvement in the performance • cross-modal real-time information available to travelers How systems will evolve: • A related development is the emergence of increasingly more sophisticated geospatial and spatio-temporal information management capabilities. These factors have the potential to revolutionize traveler services, and the provision and analysis of related information. In this revolution, travelers and sensors in the infrastructure and in vehicles will all produce a vast amount of data that could be interpreted and acted upon to produce a sea change in transportation.