An Integrated System for Perception-Driven Autonomy with Modular Robots (1709.05435v2)

Abstract: The theoretical ability of modular robots to reconfigure in response to complex tasks in a priori unknown environments has frequently been cited as an advantage and remains a major motivator for work in the field. We present a modular robot system capable of autonomously completing high-level tasks by reactively reconfiguring to meet the needs of a perceived, a priori unknown environment. The system integrates perception, high-level planning, and modular hardware, and is validated in three hardware demonstrations. Given a high-level task specification, a modular robot autonomously explores an unknown environment, decides when and how to reconfigure, and manipulates objects to complete its task. The system architecture balances distributed mechanical elements with centralized perception, planning, and control. By providing an example of how a modular robot system can be designed to leverage reactive reconfigurability in unknown environments, we have begun to lay the groundwork for modular self-reconfigurable robots to address tasks in the real world.

• The paper demonstrates an integrated modular robot system that autonomously reconfigures based on real-time sensory input and high-level planning.
• It details three hardware demonstrations where the robot adapts its configuration for tasks like object retrieval, stair climbing, and reaching elevated targets.
• The study highlights both the system’s dynamic adaptability and the potential for enhanced robustness to overcome hardware and software failures.

An Integrated System for Perception-Driven Autonomy with Modular Robots

The paper presents a sophisticated modular robotic system that integrates perception, high-level planning, and modular hardware to accomplish complex tasks in environments unknown to the system prior to operation. This work systematically explores how modular self-reconfigurable robots (MSRR) can autonomously perceive, explore, and manipulate their environment by dynamically changing their configuration based on task requirements and environmental constraints.

The central contribution of the paper is the demonstration of an MSRR system that autonomously performs high-level tasks, such as object retrieval and delivery, by exploiting its ability to reconfigure itself in response to sensory input about the environment. The system balances distributed mechanical elements with centralized perception, planning, and control to execute tasks effectively. This integration is validated through three distinct hardware demonstrations.

Key Demonstrations and Observations

1. Hardware Demonstrations: The system was tested in three scenarios:
   • Demonstration I involved exploring an environment to retrieve colored objects and deliver them to a designated drop-off zone. The robot autonomously reconfigured itself to a "Proboscis" configuration to retrieve an object from a narrow corridor.
   • Demonstration II required the robot to deliver a circuit board to a mail bin at the top of stairs. The robot utilized the "Snake" configuration to traverse the stairs and accomplish the task.
   • Demonstration III focused on placing a stamp on a package elevated above ground level, where it reconfigured into the "Proboscis" to reach the height needed.
2. System Architecture and Implementation: The system employed the SMORES-EP modular robot, using a central sensor module fitted with cameras and computation units for mapping, navigation, and environment classification. The high-level planner orchestrated robot actions based on perception-informed decisions.
3. Performance and Failures: While the demonstrations showcased successful task completions, the system also experienced failures primarily due to hardware errors and low-level software issues. It was observed that the robustness of the system could be enhanced by incorporating more feedback mechanisms and higher error tolerance in its centralized components.
4. Reconfiguration Strategy: The system's ability to reconfigure was dependent on real-time perception information processed through a centralized control paradigm. This flexibility allowed the robot to adaptively select and execute different configurations necessary for completing the specified tasks.

Implications and Future Directions

The experimental results underscore the potential of autonomous modular reconfigurable robots in performing sophisticated tasks without pre-programmed knowledge of the operational environment. This capability highlights the growing importance of integrating perception with robotic platforms to enable them to function effectively in dynamic and previously unknown settings.

Practically, this could lead to advancements in search and rescue operations, space exploration, and autonomous service robots that can adapt their physical form to meet the challenges of their environment. Theoretically, the paper provides insights into systems design for modular reconfigurability, pointing toward enhancements in distributed and centralized robotic systems.

Future research directions may focus on enhancing the robustness and fault tolerance of centralized systems and improving the flexibility of environment characterization functions to handle unstructured environments efficiently. Additionally, further work could address the optimization of energy consumption during reconfiguration processes and develop algorithms to automate reconfiguration plans.

The paper marks a step toward realizing the full potential of MSRR systems in real-world applications, setting the stage for more autonomous, adaptable, and perceptually aware robotic systems.