Gamma Two Robotics

Gamma Two's Cybernetic Brain

Whether you are developing a mobile robot designed for mining applications or a docent to show people around a new museum exhibit, if your robot is going to work in the real world, it probably needs a brain. Gamma Two Robotics has developed a powerful Cybernetic Brain that enables almost any robot to do its job more effectively. If your robot needs to talk to an audience, the Cybernetic Brain will allow it to tailor what it says to the responses of the audience, react appropriately to the crowd, and adapt the interaction with knowledge of what has happened in the past. If your robot needs to achieve mission goals in complex dynamic environments, the Cybernetic Brain will allow it to replan on the fly, adapt to changing conditions, and take advantage of opportunities as they present themselves.

The Cybernetic Brain is designed to function in the same dynamic and uncertain world that we live in. Unlike the tightly controlled world of industrial robotics, where robots can simply repeat the same actions over and over, many robotic applications require the system to interact with people and the environment in a dynamic manner. This can be done with tele-operated systems, like undersea exploration robots, or bomb disposal robots. But this requires a person in the loop, to provide the needed flexibility. The Cybernetic Brain is designed with the flexibility to handle day to day uncertainty, freeing up people to deal with the hard problems. The Cybernetic Brain is designed to meet those needs.

We have designed the brain to maintain a thin interface between the brain and the hardware that drives the robot. This integration simplifies the integration of our Cybernetic Brain with third-party robots. This gives those robots all the autonomy and intelligence needed to achieve complex tasks in the real world. The process of this integration is straight-forward, proven, and cost effective. Rather than trying to develop new controllers to achieve complex tasks, we can integrate our tested and reliable solution with your hardware.

Process

1. Requirements/Analysis
2. Interface Design
3. Knowledge Base Construction
4. System Integration/Testing

The process begins with gathering requirements - what is the robot designed to do? What are the tasks, what is the environment? Once these requirements are known, the capabilities of the robot are analyzed, and a model of the informational requirements of the brain is engineered. From these requirements, a validation test suite can be crafted, assuring that final systems meets your needs.

Next we analyze the design of the interface between the Cybernetic Brain and your robot. In many cases, the existing control interface is sufficient - after all, it was designed to enable the robot to do its job. However, in some cases, additional functionality needs to be incorporated to enable the brain to control the robot.

The next step is to construct the knowledge base that the Cybernetic Brain will use to model the robot, the environment, and the tasks and goals of the system. This knowledge base enables it to make the appropriate plans and decisions to control the robot.

The final stage is the integration of the robot, the interface, and the brain. The integrated robotics system is then fully tested to verify that the complete system meets the requirements defined in the first stage of the process. Upon completion, your system will be more capable, more adaptable, and more intelligent. If this is what you are looking for, get in touch.

Requirements/Analysis

The first, and most critical, phase of this process is gathering requirements, and analyzing the needs. You know the purpose and function of the robotic system, you know the needs of your customers. We work with you to start from your existing requirements, and evolve them into specifications for the Cybernetic Brain's Knowledge Base.

We capture a formal model of the environment in which the robot will function, the tasks it must complete, and capabilities the robot can use to achieve those goals. Lastly, we work with your engineering team to determine the available sensors and control interfaces of the robot.

Interface Design/Implementation

For the Cybernetic Brain to control the robot intelligently, it needs to have access to both sensor data and control systems. We refer to this combination as the Perception/Action component. Many robotic systems (such as tele-operated robots) provide sensor information to a human operator, who is responsible for interpreting the sensor data, and making the decisions. For some robotic systems, additional sensor data must be provided to the Cybernetic Brain, which may require additional hardware interfaces. In other cases the sensor data may be available in analog formats which must be converted to digital data for the Cybernetic Brain.

Gamma Two has partitioned the Cybernetic Brain, so that the data interface between the robotic hardware and the software is a thin as possible. This minimizes the complexity of the interface, and reduces the cost of integrating the brain with the robot. Using the model developed in the first phase of the process, we can design a complete interface description, which details the necessary sensor data, and defines an API for the robot's low level controller.

Knowledge Base Construction

The goal of the previous steps is to enable intelligent, context-aware control of the system. The Cybernetic Brain is a general purpose planning/execution system that evaluates multiple ways to achieve a goal, and selects the best solution. This requires a detailed model of the actions that the system to undertake to achieve its goals, the environment in which the actions take place, and some of the possible adverse outcomes of those actions. This model is stored in an external data base, making it easy to adapt to changing missions, and to incorporate upgrades to the robotics hardware, or changes in the environment. While this is arguably the most complex and demanding phase of the process, we have evolved this stage to a science.

Once this knowledge base is created and tested, it is possible to run detailed simulations of the behavior of the system in response to different conditions. These simulations are used to confirm the correct behavior of the system. This provides several benefits to the system, including the ability to compare different approaches to achieving the mission with out the need for extensive real world testing. Rather, options can be compared, and the highest ranked candidates are then tested in the real world, saving both time and money.

System Integration/Testing

• Interface between the robot system and the brain,
• The Knowledge Base capturing the model of the world, the robot's capabilities and missions, and
• The requirements and acceptance criteria for the integrated system.