Miss Piggy

Miss Piggy is the main robot developed by the Pequi Mecânico Robotics Team for the RoboCup@Home league, designed to assist humans in domestic environments through advanced manipulation, interaction, navigation, and perception systems.

Miss Piggy

Manipulator

Equipped with a ViperX 300s 6DoF manipulator, powered by 8 Dynamixel servos, Miss Piggy can handle up to 750g. The manipulator uses a combination of classical control and reinforcement learning algorithms for task execution.

Human Interaction

Miss Piggy interacts primarily via voice commands, using Neural Networks optimized for embedded systems. Speech recognition and synthesis are powered by the NVIDIA Riva SDK, while a custom Llama 2 model enables Natural Language Understanding (NLU) for contextual task processing. A web-based GUI allows additional control and feedback.

Computer Vision

The robot employs Yolo11-based models for object and pose detection, enhanced with TensorRT inference for near real-time performance, enabling effective environmental perception and task execution.

Navigation and Mapping

Miss Piggy uses the Navigation 2 stack for obstacle avoidance and path planning, with data from LIDAR, point clouds, and a 2D occupancy map generated through SLAM. Localization is achieved using wheel odometry, visual odometry, IMU data, and algorithms like EKF, ORB-SLAM, and AMCL.

Key Components

  • Jetson Xavier AGX: Handles AI tasks including vision and speech recognition;
  • Clearpath Jackal UGV: Provides a mobile base for autonomous navigation;
  • ECS BOX: Manages power distribution and regulation for all robot components.