Is this project an undergraduate, graduate, or faculty project?
Undergraduate
Project Type
individual
Campus
Daytona Beach
Authors' Class Standing
Karthik Kolipaka, Junior
Lead Presenter's Name
Karthik Kolipaka
Lead Presenter's College
DB College of Engineering
Faculty Mentor Name
Cagri Kilic
Abstract
Investigating the Operational Performance of Heterogeneous Multi-Robot Systems for Obstacle Pushability Affordance
Karthik Kolipaka - Embry-Riddle Aeronautical University, Daytona Beach, FL, United States
Space exploration presents unique challenges due to the complexity of space environments like the rugged terrain, extreme temperatures, and communication delays. Traditional homogeneous robotic systems, like identical robots, have limited capability to perform diverse functions for obstacle removal or avoidance under unpredictable conditions in static and dynamic contexts. Therefore, we require robotic systems that are interdependent, autonomous, resilient, and efficient in conducting joint operations for mission longevity. In such cases, Heterogeneous Multi-Robot Systems that consist of multiple non-identical robots perform various operations on planetary surfaces to support a collective objective. The goal of our research is to investigate the operational performance of heterogeneous multi-robot systems to determine the optimal point of interaction during obstacle removal and execute pushing or grasping actions based on affordance estimates. Our heterogeneous multi-robot system will consist of quadruped and multiple ground rovers that operate autonomously and share exteroceptive information and proprioceptive feedback with robotic systems to perform pushability or grasping actions. The quadruped acts as a flagship and executes the final mechanical actions based on tactile and visual data collected from other robotic systems. The quadruped and ground rovers are equipped with Thermal and LiDAR sensors to provide the 3- dimensional topological mapping of the obstacles and determine the visual aspects such as the size, shape, and surface characteristics of the object. The ground rovers also contain force or load sensors that determine the force measurements required for pushing obstacles on the planetary surface. The visual data accompanied by tactile data is then analyzed to determine the pushability vs grasping affordances via convolutional neural networks. Once the mode of action is decided, the sensor data is analyzed for optimal interaction points on the obstacle surface to execute pushability or grasping actions at these points through respective automated algorithms. The quadruped performs the decision-making analysis and commands the integrated mechanical arm to push movable objects or grasp prehensile obstacles based on the visual and tactile data analysis outcomes. From affordance estimates based on data collected from multi-robot systems, we can determine the ideal possibility for obstacle removal or avoidance and develop a framework for probabilistic affordance of an obstacle for autonomous navigation and guidance on planetary surfaces.
Did this research project receive funding support (Spark, SURF, Research Abroad, Student Internal Grants, Collaborative, Climbing, or Ignite Grants) from the Office of Undergraduate Research?
No
Heterogeneous Multi-Robot Systems for Obstacle Pushability Affordance
Investigating the Operational Performance of Heterogeneous Multi-Robot Systems for Obstacle Pushability Affordance
Karthik Kolipaka - Embry-Riddle Aeronautical University, Daytona Beach, FL, United States
Space exploration presents unique challenges due to the complexity of space environments like the rugged terrain, extreme temperatures, and communication delays. Traditional homogeneous robotic systems, like identical robots, have limited capability to perform diverse functions for obstacle removal or avoidance under unpredictable conditions in static and dynamic contexts. Therefore, we require robotic systems that are interdependent, autonomous, resilient, and efficient in conducting joint operations for mission longevity. In such cases, Heterogeneous Multi-Robot Systems that consist of multiple non-identical robots perform various operations on planetary surfaces to support a collective objective. The goal of our research is to investigate the operational performance of heterogeneous multi-robot systems to determine the optimal point of interaction during obstacle removal and execute pushing or grasping actions based on affordance estimates. Our heterogeneous multi-robot system will consist of quadruped and multiple ground rovers that operate autonomously and share exteroceptive information and proprioceptive feedback with robotic systems to perform pushability or grasping actions. The quadruped acts as a flagship and executes the final mechanical actions based on tactile and visual data collected from other robotic systems. The quadruped and ground rovers are equipped with Thermal and LiDAR sensors to provide the 3- dimensional topological mapping of the obstacles and determine the visual aspects such as the size, shape, and surface characteristics of the object. The ground rovers also contain force or load sensors that determine the force measurements required for pushing obstacles on the planetary surface. The visual data accompanied by tactile data is then analyzed to determine the pushability vs grasping affordances via convolutional neural networks. Once the mode of action is decided, the sensor data is analyzed for optimal interaction points on the obstacle surface to execute pushability or grasping actions at these points through respective automated algorithms. The quadruped performs the decision-making analysis and commands the integrated mechanical arm to push movable objects or grasp prehensile obstacles based on the visual and tactile data analysis outcomes. From affordance estimates based on data collected from multi-robot systems, we can determine the ideal possibility for obstacle removal or avoidance and develop a framework for probabilistic affordance of an obstacle for autonomous navigation and guidance on planetary surfaces.