ViCAL : Virtual Cities and Autonomous Learning

Created by CGA Simulation and PINTEL, ViCAL integrates simulated data for training the deep learning algorithms behind the AI that analyses video and other sensor data from traffic control systems.

Korean based company, PINTEL, has an existing product called Precise Video Analytics Experience (PreVAX), already in use in Korea. PreVAX is an intelligent traffic system that collects real-time data by applying an artificial intelligence-based algorithm to video from traffic control systems. CGA Simulation has developed ALEAD, a simulation platform, previously optimised for training Connected and Autonomous Vehicles (CAVs), planning installation of 5G infrastructure and transport infrastructure planning.

ALEAD uses agent based modelling (ABM), which generates 'patterns-of-life' created via the interaction of digital agents (simulated vehicles), which travel from home, to work, to leisure, to retail, in manners informed by real life. This is better for predicting reactions to unprecedented dilemmas, like the introduction of new modes of travel or a pandemic.

ViCAL will add the simulation techniques developed by CGA to the AI video platform developed by PINTEL. This newly enhanced product will be a substantial revenue earner for both companies as this innovation adds unique features that distinguish them in the Smart City, V2X, traffic management and CAV markets

CGA provides synthetic data that can be combined with PINTEL's technology to predict edge cases. Using CGA's complex simulated environment, PINTEL will generate synthetic camera feeds. The simulation will visualise and remove an accessibility barrier to scenes that in real life would be dangerous or difficult to generate.

CGA can create a broad range of simulated environments. During ViCAL we will label simulated data for easier use and transference across platforms, localities, and technologies. Our created cases can then be applied to multiple scenarios and variables.

One example would be a car pulling out unexpectedly, but with a range of delays, from very close, to collision, to further away. Ease of access and clear data visualisation means ViCAL can be used by anyone, not just specialists, and shared easily with wider teams and stakeholders. CGA's hazard editor allows users to easily create and share specific driving situations for either pass/fail test of AI or equally to create a range of edge cases, with a customisable capability.