Robot Learning and Robot Behavior Control Software
Commercial aspects

Q: How much does a Skilligent license cost?
A: For pricing information, please contact Skilligent.

Q: Is it an open source product?
A: No, it is not. Please contact us if you absolutely must have the source code.

Q: Are you willing to use a software escrow trustee?
A: Yes.

Q: Does Skilligent use any third-party software such as ... (face recognition, speech recognition, tracking, etc)?
A: No. All the required software is built by Skilligent and comes pre-packaged with the software.

Q: Does Skilligent own any patents?
A: We are hoping to secure several patents soon. The work is in progress.

Algorithms

Q: Is Skilligent based of neural network algorithms?
A: No. Neural networks are a useful design pattern for certain types of tasks. Skilligent uses "traditional" software designs for the components that perform those tasks.

Q: Does it use SIFT image feature detector?
A: No, the computer vision system uses a modified Harris Corner Detector algorithm.

Q: Does Skilligent utilize any of the reinforcement learning algorithms (Q-Learning, TD-Learning ...)?
A: Yes. Skilligent utilizes some special forms of reinforcement learning in the processing and decision making hierarchy. The reinforcement learning algorithms are combined with a set of supervised learning algorithms to create a system which can acquire new skills pretty quickly.

Q: Is Skilligent software makes robots conscious?
A: No, the software is not conscious. The future versions of the software will feature some elements of consciousness; that is an area of special interest for Skilligent team.

Applications

Q: Can it learn how to cook?
A: Yes, it should be able to learn how to cook a particular dish in a particular structured environement (kitchen). Various limitations can be overcome by environmental scaffolding and other methods.

System integration

Q: Can Skilligent use my sensor (FireWire camera, LADAR, GPS)?
A: In general, Skilligent can utilize any sensors that an underlying platform provides. The trickier question is how to convert output of those sensors into a form which the behavior system can handle. Some of the sensors (such as touch sensor or a GPS sensor) don't require any pre-processing and can be used by the system directly through simple drivers. Other sensors (such as a video camera or a microphone) require intensive preprocessing. The product comes with sensor processing layers for a few kinds of sensors (video and audio). The others can be added by a system integrator using Skilligent Development Kit.