This lecture discusses the building blocks of Nural Network (NN) controller "brains", of future autonomous products, towards the level of Technological Singularity.   

This lecture highlights the differences between Supervised Learning (SL), Unsupervised Learning (UL), and Reinforcement Learning (RL) Machine Learning (ML) and why RL is the most suitable for Autonomous Control.

PID (Proportional, Integral, Derivative) controller has been the most popular motion control technology for over 100 years. This article demonstrates a design tool, which was developed by the author, to help control engineers visualizing the effect of their 3-parameter choice on robotics performance. RL, NN, has hundreds of parameters.

In this lecture we present the block diagram of RL motion controller including State, Action and Reward, with an sample of the NN training procedure.

In this lesson we take a look how different is the training process of PID and RL motion controller and how similar they in operation are after training.

If you are an engineer or scientist with interest in becoming a robotics, AI/ML, product development, expert, this lecture demonstrates the type of work you need to learn, experience to gain, and the options in getting there.

One of the most important elements of developing autonomous RL control system, is the "Reward" function. It defines the desired performance that will optimize the application objectives. In this lecture we will take a look how AWS is training the world RL Reward Functions, using DeepRacer cars with lots of competitive fun.

In this lesson we will take a closer look at RL components. It includes Agents, which manage the training optimization process. Actors which estimate for each initial random State, (i.e. the feedback from System and Environment sensors), the set of optimal Action (i.e. the command to all system motion components). And Critics, which estimate the Max Reward that the autonomous system can reach based on previous Actor Actions.

In this lecture we look at the mathematical relationship between Critic, Actor and Hyper Parameters, which is a complex nonlinear environment in which we are looking for the extreme Reward peaks. The mathematical formulation is known as Bellman equation, which is the highlight of RL process.

In this lesson we will highlight the key $B companies, which provide AI/ML solutions. Including, clouds, ML tools, data storage, and global IoT communication environment, including free training courses. Their ultimate goal is to Deploy the trained NN into, CPU / GPU control processors of autonomous systems in robotics, avionics, naval, healthcare, military, manufacturing, agriculture et al.

In this final lesson we take a look how you could get started in AI/ML Business Development (BD). The recommended process is similar to what Skeletal Axes Assistive Robots (https://saar-inc.com) has been doing with its strategic BD mission to assist global handicapped people by using SL labeled datasets and RL Rewarded agents in medical exoskeletons.