DEEP-RL BOOSTS ITS
Using deep reinforcement learning framework, in-house probabilistic modeling and inference engines, AiGENT-TECH build a large scale 3rd generation on-demand door-to-door multimodal public transport platform, supporting the on-going 'smart-rail transportation' revolution.
Running on a powerful processing power platform, we provides:
Long/short distance managed rides for mass transport systems
The most optimized rail-centric on-demand door-to-door services
A winning strategy for railway service providers to innovate and build customers loyalty
MACHINES' REAL-WORLD REPRESENTATION
AiGENT-TECH architecture uses a combination of model based augmented with model free deep reinforcement learning techniques in order to maximize profits, minimize OPEX and provide most effective end-customer experience.
Leveraging human experts know how, we developed dynamic & statistical models for urban cities.
These human-expert 'handmade' models, powered by real-time samples, let our on-demand public transport platform excel vs. any other big data solution.
WHEN TECHNOLOGY GAINS PROFITS
AiGENT-TECH deep reinforcement learning reward functions take into account all variables such as demands prediction, routes optimization, charging policy, ..., in order to take the best next action for a given situation (state).
On top of that, our cognitive algorithms support real-time constraints and realistic processing power.
The following videos, reflecting AiGENT-TECH reinforcement learning technology, show fleet level routes planning AiGENT thinking (reasoning) and planning his best next action (next move) under uncertainty and as a function of demand prediction, incoming urban traffic observations and customers dynamic scheduling providing most effective customer experience following his multi dimensions reward function.
UNDERSTAND THE IMPACT OF EVERY DECISION ON BOTTOM-LINE PROFITS
AiGENT-TECH multi-dimensions reward function lets service providers test (off-line) and understand the impact of their decisions on bottom-line profits.
Combining trains schedule, vans size, number of vans, working hours, ... in a single mathematical expression may lead to a winning strategy vs. any other on-demand door to door service in the same geographical area.