I suggest you made your platform first in simulation application on you your computer , and you can which app you are familair with like MATLAB which offers a lot of choices. and then find the spicific operation practicale platform in your lab to implement your target. First you should what the others did befor and then you can make some improvments by yourself to acheive a contrbution you looking for... good luck

Making the robot manipulator suitable for unstructured environments is an important part of the process of using the robot for tasks such as picking an object. Here are several strategies and technologies that can be employed to enhance a robot manipulator's adaptability:

Sensory Perception:

Vision Systems:

The robot should be fitted with complex visioning devices e.g. cameras/depth sensors like LiDAR or Kinetic. It enables the robot to determine the objects, obstacles, and the general scene.

Tactile and Force Sensors:

Incorporate tactile and force sensors onto the robot’s end effectors or grippers. This makes it possible for the robot to feel what kinds of characteristics things have like, shape, rigidity, and texture.

Machine Learning and AI:

Object Recognition:

Machine learning will be used to train the robot to identify and categorize objects on a real-time basis. This gives the robot a versatile nature to accommodate different object shapes.

Adaptive Planning:

Develop dynamic trajectories and grasping strategies for the robot’s movement, which will be implemented based on information received from the environment and the object properties.

Collision Avoidance:

Obstacle Detection and Avoidance:

Develop algorithms for detecting environmental obstacles and computing collision-free path planner; The presence of obstacles makes this important as it aids navigation within the crowded area to avoid any collision.

Flexible Manipulation:

Compliant Actuators:

Deploy actuators that move comfortably with non-conforming objects on the system. The robots are thus able to comply easily with the shape of the objects they interchange with to minimize damage during impact against them.

Soft Grippers:

Use soft grippers or adaptive grippers which will be able to grip any object. They tend to be far better at handling uneven objects.

Learning from Demonstration:

Programming by Demonstration (PbD):

Demonstrate and teach robots through training in the environment. With such demonstrated examples, the robot can generalize on its actions to accommodate changes in the environment.

Simulations and Virtual Environments:

Simulation-Based Training:

Train the robot using a simulated environment with various scenarios before deployment in the real world. This can also allow the robot to learn faster.

Hybrid Systems:

Wheeled or Tracked Mobility:

Add another wheeled base with the manipulator to enhance mobility, especially from areas of mixed terrain.

Modular Robotics:

Make the robot a modular system so that it can be tailor-made for a particular function.

Human-Robot Collaboration:

Collaborative Robots (Cobots):

Facilitate teamwork between robots and people. Such can also be in the form of shared control whereby the robot and human cooperate to succeed in an objective, taking advantage of each other’s strengths.

.

.

.

Identify a particular operation practice platform in your lab where you will apply the above. first, try to follow a similar approach that has been applied before them. With this in mind, you will be able to formulate your improvement measures toward achieving the contribution you look forward to making... best of luck