The topic of this thesis is about design and implementation of a model based architecture for cobotic cells. With the advent of tactile internet, regularizing coexistence of robots and humans has become imperative, meaning the so called "Cobots" need a new use case architecture for its unit cell to operate safely alongside humans and real world objects and obstacles. This architecture is based on multiple models each describing one aspect of use case aiding in functionality of cobots. For this the thesis described three models namely world model, application model and safety model which are described using different notations.\\
The world model is a global model describing the cobot and other things in its environment, giving "on the whole" information about the components in real world a cobot has, this includes one or more humans who can be moving in and out of cobotic world zone, then some obstacles and grasp object which can be a ball or cube.\\
The application model describes the flow of individual actions of grasping that can be performed by Cobot according to a motion trajectory to accomplish the given task. This model is all about performing the task and action of the cobot. Lastly, the safety model shows how a Cobot achieves goal of not causing any harm to humans or other objects in its proximity and how to respond to them by moving around them appropriately in cases imminent collisions are detected . \\
The real life problem scenario can be described as follows. Robot is expected to perform some job and to make it to do that with safety i.e. detect and evade obstacles / humans, this safety and application can be achieved in two different step cases. The models designed and described ,address to this task or problem of first, to train the robot for performing actions according to a preconceived plan using inbuilt “teaching” feature of robot and then doing it safely in real world conditions.\\
The use case can be understood by seeing a scenario where we can train the robot in a laboratory / ideal conditions and give a working functionality to it by giving a design which shows how to perform a task which robot can use to work accordingly and this is known as application model implementation. The Franka Panda robot has a teaching mode where we can set a series of poses and grasp actions manually which can train the robot to perform a task according to a plan and this can be done repetitively by the robot later in scenario 2 which is real world and has added conditions of realism.\\
For this ,complex condition are added to application model about how to respond when it detects a human in proximity and obstacles in trajectory paths and in addition this real world simulation adds real world conditions like adding torque to joints as is in real world to see if arm can for example really life an object.
\newpage
In scenario one the architecture of robot`s world model is already known and has thus been used ,its teaching capability to train it to move to a coordinate position and then start a trajectory for instance at position X to move a position close to an object that is needed to be say picked up and then it can use its gripper to pickup the object and again move arm to another desired location where it want to drop the object and there it releases the gripper to put that object down and thus completing the task at position Y. This is part of Application Model as described before.\\
This is smaller use case replication of saying a robot actually moved but here the idea is restricted to only moving arm which is the same when it comes to functionality achieved by robot moving itself vs moving its arm as previously mentioned, and this is fulfilling the same work of detecting things in proximity and achieving the tasks by completing trajectory as well as at same time to do it safely by responding appropriately as per intended use case programmed for safety.\\
So far above description talks about training the robot in scenario 1 and now another scenario is considered which is a real world task where the robot is made to perform the same work it was trained in Scenario 1 but in real life and this means the safety aspect should now be built into the scenario and for this a safety architecture is constructed which is used by robot, by telling it how to respond when seeing an obstacle like a cube or box for example or a human being.\\
