Using Explainable AI and Hierarchical Planning for Outreach with Robots
Project Overview
The document discusses the creation and utilization of JEDAI.Ed, an innovative open-source educational platform aimed at teaching robotics programming and artificial intelligence concepts to K-12 students. By providing an intuitive user interface and adaptive curriculum generation, JEDAI.Ed simplifies the complexities of robot planning and incorporates natural language explanations to facilitate learning. User studies reveal that JEDAI.Ed markedly improves student engagement, comprehension, and task-solving efficiency when compared to its earlier version, JEDAI. Through these enhancements, the platform demonstrates the potential of generative AI in education, showcasing how tailored learning experiences can transform traditional teaching methodologies and better equip students with vital skills in technology and programming.
Key Applications
JEDAI.Ed - an open-source platform for robotics programming education
Context: K-12 educational settings, particularly for high school students with little to no computer science background
Implementation: The platform uses a visual interface with block-based programming (Blockly), an adaptive curriculum generation method, and incorporates explainable AI principles to guide users through robotics programming.
Outcomes: Increased student engagement and interest in AI, improved understanding of robotics programming concepts, decreased time needed to solve tasks, and positive feedback from users.
Challenges: Initial technical knowledge barriers for users, network latency issues during pilot programs, and the need for continual improvements in accessibility for non-keyboard/mouse devices.
Implementation Barriers
Technical
Network latency issues during the pilot program affected the user experience, particularly in rendering video streams.
Proposed Solutions: Optimizing the motion planner to handle trajectory packets more efficiently, and plans to improve accessibility for devices without keyboards/mice.
Usability
Some users found the initial instructions unclear, particularly regarding the starting block in the interface.
Proposed Solutions: Feedback suggested renaming the start block to 'Connect blocks here' to enhance intuitiveness.
Project Team
Rushang Karia
Researcher
Jayesh Nagpal
Researcher
Daksh Dobhal
Researcher
Pulkit Verma
Researcher
Rashmeet Kaur Nayyar
Researcher
Naman Shah
Researcher
Siddharth Srivastava
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Rushang Karia, Jayesh Nagpal, Daksh Dobhal, Pulkit Verma, Rashmeet Kaur Nayyar, Naman Shah, Siddharth Srivastava
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai