How rohanaaron/portfolio works
This is a robotics and applied-ML project portfolio, not one integrated product. Its clearest runtime workflow is the Bayesian IRL demo: a robot sorts objects, a human or simulated human corrects a placement, and the learner updates its inferred preference while the simulation displays progress. The repository also documents iterative 3D refinement and factor-graph SLAM as separate project workflows. Edges reflect the shown main-project imports and explicit README descriptions; unsampled project implementations are not inferred as connected.
Main entry points: main_project.py Sorting Demo · text_to_text_entropy_ratio_FIXED (1).py Entropy Feedback · final_script_FIXED_v2 (1).py 3D Generation · Semantic_score_VLM_VQA (2).py VLM Scoring
Components 16
3D Refinement
- Text Prompt
graduate/CUBIST/example_text (2).py - 3D Generation
graduate/CUBIST/final_script_FIXED_v2 (1).py - VLM Scoring
graduate/CUBIST/Semantic_score_VLM_VQA (2).py - Entropy Feedback
graduate/CUBIST/text_to_text_entropy_ratio_FIXED (1).py - Prompt Boosting
graduate/CUBIST/bigram_boost_FIXED (1).py - Prompt Filtering
graduate/CUBIST/prompt_filter (2).py
Human Intent Learning
- Sorting Demo
graduate/projects/Bayesian IRL Human Corrections/main_project.py - Objects and Bins
graduate/projects/Bayesian IRL Human Corrections/sort_objects_project.py - Panda Arm
graduate/projects/Bayesian IRL Human Corrections/robot.py - Human Corrections
graduate/projects/Bayesian IRL Human Corrections/human_model_project.py - Bayesian IRL
graduate/projects/Bayesian IRL Human Corrections/bayesian_irl_project.py - Learning Display
graduate/projects/Bayesian IRL Human Corrections/display_project.py
Factor-Graph SLAM
- Marker Capture
graduate/projects/ArUco-SLAM-Factor-Graph/controller.py - Factor Optimizer
graduate/projects/ArUco-SLAM-Factor-Graph/A4.py
Archive Transcription
- Historical HTR
README.md
Other components
- Human Operator
Connections 13
- Text Prompt → 3D Generation: guides generation
README.md - 3D Generation → VLM Scoring: evaluated by
README.md - VLM Scoring → Entropy Feedback: scores feedback
README.md - Entropy Feedback → Prompt Boosting: refines prompt
- Entropy Feedback → Prompt Filtering: filters tokens
- Human Operator → Sorting Demo: starts demo
- Sorting Demo → Objects and Bins: uses scene
graduate/projects/Bayesian IRL Human Corrections/main_project.py - Sorting Demo → Panda Arm: controls arm
graduate/projects/Bayesian IRL Human Corrections/main_project.py - Sorting Demo → Human Corrections: requests correction
graduate/projects/Bayesian IRL Human Corrections/main_project.py - Sorting Demo → Bayesian IRL: updates preferences
graduate/projects/Bayesian IRL Human Corrections/main_project.py - Sorting Demo → Learning Display: updates display
graduate/projects/Bayesian IRL Human Corrections/main_project.py - Bayesian IRL → Objects and Bins: scores bins
graduate/projects/Bayesian IRL Human Corrections/bayesian_irl_project.py - Marker Capture → Factor Optimizer: feeds landmarks
README.md