Autonomous vehicle: two computational approaches that transform this industry.

Alongside to the new IT Convergence, the self-driving vehicle’s trend emerges with great expectations. We are still at the early stage of this trend, but it is exciting to see that, self-driving can improve our security, can optimize our outcome combined with the comfort.

Google car among others prestigious projects can transform the way we move, travel and transport. 

Connectikpeople.co, soon #Retinknow® recalls that, planning an autonomous vehicle’s course often involves an approach called Markov Decision Process (MDP), a sequential decision-making framework.

We can learn that, in this schema (“tree” of possible actions), each node along a tree can branch into several potential actions; each of which, if taken, may result in even more possibilities. 

Agha-mohammadi explains it, as “the process of reasoning about the future” to determine the best sequence of policies to minimize risk.

If unfamiliar, Markov Decision Process (MDP), works rationally well in environments with perfect measurements, where the result of one action will be observed perfectly. But in case of uncertainties in measurements, such sequential reasoning is catastrophic.

To solve this issue, a general framework of Partially Observable Markov Decision Processes (POMDP) can be crucial.  

Connectikpeople.co, soon #Retinknow® can observe that, this approach can generate a similar tree of possibilities, although each node represents a probability distribution, or the likelihood of a given outcome: funnel multiple possible outcomes into a few most-likely outcomes.

This means the combination of Partially Observable Markov Decision Processes (POMDP) and the Markov Decision Process (MDP), shutdowns uncertainties and brings safe, accuracy, predictions and performance.