World Car-Free Day, celebrated annually on 22^nd of September, is an initiative that promotes the notion of car-light societies and encourages citizens around the world to rethink their dependence on private cars for transportation. With 20% of the world’s carbon emission currently being transport-related and this number continuing to grow, any step towards better efficiency holds great promise in solving this global challenge. While the focus on long-term behavioural change is an important step in creating and sustaining a greener world, it must be accompanied by improvements in existing transport infrastructure and technology. 

 
Pradeep Reddy Varakantham, Professor of Computer Science at SMU is contributing to this space by introducing a new approach in the way on-demand ride-pooling apps (e.g. GrabShare and UberPool) are programmed. By designing an algorithm that factors in the ‘future impact’ of each customer-to-vehicle assignment, Prof Varakantham and his team are aiming to improve the efficiency of ride-pooling systems for the benefit of the environment, customers, drivers and the aggregation companies themselves. 

Improving ride-pooling algorithms 

Prof Varakantham shares that existing approaches used by ride-pooling companies focus on assigning customers to the nearest car, an approach that he considers to be “very myopic” because it does not consider the knock-on effects. He says, “If you assign a request which is going to a very remote location, where it is hard to find subsequent requests for the driver, then that results in a loss of revenue. Instead, if there is someone who is going there to end their shift, then that's probably a better idea.” 

To do so, the programming algorithm needs to be able to think a few steps ahead and this is where Prof Varakantham and his team are breaking new ground. He says “with our approach, we not only take into consideration how close a driver is to the customer’s pick-up point, but also where serving this request will take the driver to. If serving the request takes the driver to an office district where demand is high, then that would be considered a better assignment since it is not only helping with the current request, but also plans for multiple requests in the future”. Termed the HIVES approach due to the use of a novel HIerarchical ValuE decompoSition (HIVES) framework, this approach builds on the myopic approaches typically employed by ride sharing companies.  

How do ride-pooling apps work?

Most ride-pooling services utilise the myopic approach which previously provided the best results for the ride-pool matching problem (RMP). In the figure below, steps A, B, D and F are used by existing approaches. Professor Varakantham’s HIVES approach builds on this by factoring in the ‘future impacts’ of the assignment. It does so by utilising: 

1. Step C to evaluate the value of an assignment; and 

2. Step E to train the estimates on values of an assignment. 

• Step A: Ride-pool request is received: The programme approximates each driver to its nearest road intersections. The blue dotted line in the figure denotes the path from the pickup location to the drop off location of the request. The two triangles denote the existing pick up/drop off locations of the vehicle while the black/grey dotted line denotes their current trajectory. 
• Step B: Feasible actions are generated: The programme maps the combination of requests and available drivers, and generates feasible actions for each vehicle.   
• Step C: Feasible actions are scored: This step (not used in traditional approaches) is part of the HIVES methodology and involves the scoring of all feasible actions using individual neural networks that process the different sets of data. 
• Step D:  Assignment of driver to customer: Assignment is done using an Integer Linear Programme (ILP), based on scores from Step C.  
• Step E: Value function is updated for future decision-making: As part of the HIVES approach, the programme uses the best action recommended by the ILP to update the neural network based individual value functions.  
• Step F: Execution of assignment: The drivers move to execute the ride pool assignment. 

Benefits to the environment 

To test the true effectiveness of the approach, Prof Varakantham and his team conducted a simulation and comparative study using a real-world city scale data set. A key question they hoped to answer was whether the HIVES approach could lead to real benefits for the environment by reducing the number of cars needed and the distances travelled. The results proved to be very promising, with a 9.7% reduction in the number of vehicles needed to serve the same demand. 

The study also found that the total distance covered per driver per day, was more than 10% less than the baseline. This, by extension, would translate to lower fuel consumption and carbon emission.  

Adopting a mindset of ‘thinking ahead’ 

For Prof Varakantham, this approach of ‘thinking ahead’ is critical, and offers a win-win situation for all stakeholders. He shares that beyond the environmental benefits of the HIVES approach, drivers and ride-pooling companies also stand to benefit from higher revenues when routes are better optimised. His contribution to the space sets an example for other programmers, as well as business leaders and governments, by encouraging them to reassess whether ‘future impacts’ are sufficiently taken into account in their daily operations and decision-making.  
 
A major mindset shift is needed if countries are to reach their goal of reducing carbon emission by 45% by 2030, as per the 2016 Paris Agreement. Asia was found to be the largest emitter of transport-related carbon in the year 2019.  Developing countries, in particular, have become a cause for concern since they lack the resources to transition to greener transport alternatives. By making the right adjustments to their products, services and processes, businesses and other organisations can make a marked difference in reducing global carbon emission.  Prof Varakantham advises leaders to view their organisation as a chessboard and always think a few steps ahead; with the right move, powerful outcomes can be achieved.