Preventing Collisions! Leveraging AI for Anticipating Car Crashes and Enhancing Road Safety

**Improving Road Safety with Predictive Analytics: Insights from Hari Balakrishnan**

*Understanding the Increasing Road Fatality Problem*

Hari Balakrishnan, an expert in the field of road safety, has expressed concerns about the rising number of road fatalities and proposes solutions to enhance outcomes. Interestingly, his insights may hold even more value in the era of self-driving cars.

In the United States, road fatalities have seen a significant increase of 21% since the onset of the pandemic, with an overall rate of 1.3 fatalities per 100 million miles traveled. Balakrishnan identifies three key components contributing to road risk: driver risk, vehicle risk, and road risk. While stakeholders have a good understanding of driver and vehicle risk, he emphasizes that road risk, pertaining to infrastructure and other contextual factors, remains less well-known.

**The Elusive Nature of Road Risk**

Balakrishnan highlights the challenge of quantifying road risk, which arises from factors like the road geometry, topology, and the behavior of other drivers. While advancements in technology have allowed for better data collection and evaluation, accurately quantifying road risk remains a complex task.

To address this challenge, Balakrishnan proposes the use of high-resolution crash maps to aggregate data and predict areas with a higher likelihood of crashes. However, he acknowledges that obtaining crash rates can be difficult due to the infrequency of accidents. Only about 20% of severe crashes occur repeatedly in the same location, posing a data discovery problem.

**Building a Comprehensive Understanding through Data-Driven Models**

To overcome the data scarcity issue, Balakrishnan suggests leveraging components such as geometry imaging and historic pressure information. By employing a “residual network,” a combination of various data sources, researchers can develop a comprehensive understanding of the factors leading to crashes.

The resulting data and inferences can be validated against future crash data, enabling researchers to identify risky areas. This two-point process involves pinpointing high-risk zones and implementing safety measures like safe routing to mitigate crashes.

Moreover, Balakrishnan emphasizes the importance of personalizing risk models. With the advent of the connected vehicle revolution, stakeholders aspire to extend data discovery beyond automobiles to include other modes of transport, such as bicycles and pedestrians.

**Extending Safety Measures to Pedestrians and Bicyclists**

Acknowledging the alarming 77% increase in pedestrian deaths in the last decade, Balakrishnan emphasizes the need to extend safety measures to include vulnerable road users. He attributes these fatalities to distractions caused by smartphones and advocates for proactive solutions.

To address this, Balakrishnan discusses a developing LIDAR project aimed at measuring the volume, mass, proximity, and speed of vehicles to provide advanced warnings to bicyclists and pedestrians. He envisions a future where individuals wear jackets or clothing embedded with sensors to enhance road safety for all.

**About Hari Balakrishnan**

Hari Balakrishnan holds the Fujitsu Professorship in the EECS Department at the Massachusetts Institute of Technology (MIT). He leads the Networks and Mobile Systems group at CSAIL and is the founder and CTO of Cambridge Mobile Telematics. Moreover, he co-founded StreamBase Systems and has received numerous accolades, including the ACM SIGCOMM award for lifetime contributions.

In conclusion, Hari Balakrishnan’s insights on predictive analytics and road safety offer valuable perspectives in addressing the increasing number of road fatalities. By leveraging data-driven models, high-resolution crash maps, and personalized risk models, stakeholders can make informed decisions to mitigate road risks and create safer environments for all road users.