Real-time intent prediction of pedestrians for Autonomous Vehicles



Autonomous Vehicle testing in a civic atmosphere has become regular scenery across California, Detroit, Arizona, London, Beijing and diverse parts of the world.

The tests indicate that autonomous technology is getting promoted from a long distance thought to near future action. In the near future, Autonomous vehicle will share the road with:

  1. Manual automobiles (Bus/Cars/ Trucks)

  2. Other autonomous vehicles (Bus/Cars/ Trucks)

  3. Pedestrians

No matter how extreme the autonomy and new complementary technologies get advanced, the safety of pedestrians on the road can’t be disregarded as a solved problem. The improvements in platoon technology, V2X, and connectivity technologies have enhanced the way in which vehicles communicate with each other and how they coexist. Technology is letting vehicle to interact with the other highway users, though, pedestrian interaction with the vehicle still looks to be a grey field of advancement.

Innovations to follow!


Pedestrian intentions (in general Human intentions) are considered fickle. Pedestrian detection algorithms are prominent factors of mobile robots, such as autonomous vehicles, which directly link to human safety. Performance disparities in these algorithms could convert into disparate impact in the form of biased accident events. Innovations in the field of AI are now letting autonomous vehicles to perceive the dynamic pedestrian intent.


The scholars from Deakin University, Australia have shown up with a real-time Intent Prediction of pedestrians for Autonomous Ground Vehicles via Spatio-Temporal DenseNet which outperformed other standard procedures. The researchers at the University of Michigan have suggested a revised algorithm for envisioning the acts of pedestrians that take into account not just what they’re doing, but how they’re doing for better vehicle-pedestrian interaction in self-driving vehicles. In the system, data collected by vehicles through cameras, LiDAR and GPS allowed the researchers to capture video snippets of humans in motion and then recreate them in a 3D computer simulation. With that, they’ve created a “biomechanically inspired recurrent neural network” that catalogs human movements.


How startups are disrupting vehicle-pedestrian interactions?

Last March 2018 has witnessed Uber’s fatal crash involving a pedestrian crossing the road. The examination of the accident pointed out the ‘false positive’ generated in Uber’s program. To solve problems on this front, startups are developing advanced technologies that understand pedestrian intents.


Perceptive Automata is a startup working on human behavior prediction technology for the safe large-scale rollout of highly automated (L2/3) and autonomous (L4/5) vehicles, especially in civil localities. Perceptive Automata is functioning with OEMs, suppliers, and tech firms that are establishing or integrating ADAS and autonomous driving systems.

Humanizing Autonomy is developing a natural synergy between autonomous vehicles and humans. They develop human-centered solutions for human-machine interactions, using machine vision. Its proof of concept allowed pedestrians to communicate with autonomous cars through machine-learned gestures. The company is working with mobility providers in Europe, the U.S., and Japan, including Daimler Mercedes Benz and Airbus.

INTVO is a startup focusing on artificial intelligence for vision with the intention of reinforcing the safety and comfort of people. The company develops artificial intelligence that predicts pedestrian behaviors. Also, simulation startups like Applied Intuition validate autonomous software with human behavior models, to train AV software stack.

Industry activity 2019!

  • Volvo Trucks North America has partnered with Perceptive Automata on a project that leverages human intuition AI utilizing that reads the intention and awareness of vulnerable road users such as pedestrians, cyclists, and motorists, to enhance the situational awareness of truck drivers.

  • Humanizing Autonomy, developer of intent prediction platform that predicts the full range of pedestrian behavior across different environments and cities has risen seed funding led by Anthemis Group, the leading global Fintech, and Insurtech investor, with Japanese VC fund Global Brain Corporation, Germany-based Amplifier, and Silicon Valley-based Synapse Partners.

To conclude..


Pedestrian safety should not be compromised at any level. There is a necessity of technology that is able to predict the full range of pedestrian intention and vulnerable road user behavior in real time for safer, and more trustworthy mobility systems. The startups developing solutions will find immense opportunities in real-world implementation of automated vehicle technologies. Human intuition understanding also helps in industrial application where humans and machines co-exist.


To deep dive and stay continuously updated about the most recent global innovations in Autonomous Vehicles and learn more about applications in your industry, test drive WhatNext now!


Image Courtesy : www.humanisingautonomy.com

5 Chancery Lane, London, EC4A 1BL, United Kingdom

Priva

©2020 by WhatNext Global plc.

  • LinkedIn
  • Twitter
  • YouTube