DISHITA TURAKHIA, Ph.D.

---

This VR training simulation prototype presents an open-ended approach to practice drilling using a 3-axis milling machine. The simulation employs near “end-to-end” instruction through a safety module, a setup and drilling tutorial, and open-ended practice complete with warnings of mistakes and failures.

(published at IEEE ISMAR’23 and Demo at ACM CHI'24)

 DOI | PDF 

Our findings from a benchmarking study of 19 industry applications reveal that current systems are moderately open-ended for goal selection and system adaptability, but have significant limitations in open-endedness for product creation, feedback mechanisms, and learning paths.  

(published at IEEE ISMAR EA'24)

 DOI | PDF

We explore the design of an ACT-R based ITS framework for VR training of manufacturing skills like machining, drilling, or lathing. We detail the framework components which include designing instructions, constraint-based system model, warning and error states, action-based and time-based feedback, and the ITS architecture.

(published at IEEE ISMAR EA'24)

 DOI | PDF

The Satori system combines the users’ self knowledge and task goal of the immediate user action with a large-language model (LLM) model to provide visual assistance relevant to the user’s immediate needs.

(published at ACM CHI'25)

 DOI | PDF

InsightAR is an LLM-based visual analytics tool to integrate event detection in egocentric AR task videos and extract meaningful task insights, and to refine summaries, compare  performances, and generate actionable feedback.

(under review)

This workshop  was a collaborative exploration of cutting-edge technologies and methodologies for capturing, replaying, and sharing memory-rich experiences by bringing together experts and innovators across diverse disciplines.

(published at IEEE VR EA'25)

 DOI | PDF

A key innovation of this work was leveraging the shape-changing abilities of interactive tools to adapt the training task difficulty. User studies validating the impact of this approach demonstrated that adaptive learning not only outperformed non-adaptive training but also surpassed manually-adaptive training

(published at  ACM TEI'21)

DOI | PDF | Video | Project | Talk | MIT News

The Adapt2Learn toolkit expanded the design space for adaptivity, by empowering designers to build adaptive tools for varied skills. Adapt2Learn generates customized learning algorithms for their tools and provides a visualization feature for evaluating if the tool adapted correctly to the learner's performance.

(published at ACM DIS'21)

 DOI | PDF | Project | Video | Talk 

This project, in collaboration with the Meta Reality Labs, studies the design space of AR for adaptive skill learning. Our studies on using adaptive AR tools for training shed light on which visual & environmental cues the learners rely on for learning. The findings unveiled new design opportunities to adapt virtual cues for optimal personalized training in AR.

(accepted at ACM GI'25)

Reflection allows learners to analyze their actions and gain a deeper understanding of the skill and improve learning outcomes. The Reflective Maker - a toolkit for educators to design reflection exercises for novice makers, leverages Donald Schon's framework of reflection in making for teaching makerskills.   

(published at ACM UIST'22)

 DOI | PDF

To support the self-learning of makerskills among individuals who rely on online tutorials like Instructables, Reflectables explores the use of Large Language Models (LLMs) for generating personalized tutorials. Preliminary studies show increased self-efficacy and knowledge of tools among learners.

(under review)

Reflective Make-AR is an augmented reality system for monitoring, prompting, and recording the learners' “in-action reflections.” Our studies on the impact of immersive multimodal reflection on the learning experiences of makers indicated the need to adopt a “less is more” approach when designing interventions

(published at ACM CHI ’23)

DOI | PDF

This work studies how educators scaffold the learning of makerskills, through in-depth interviews with educators in makerspaces across Greater Boston and global fab labs. This study validated how educators prioritize a myriad of learning goals beyond mere tool expertise, such as enhancing self-efficacy and creativity.  

(published in ETRD Journal'23)

 DOI | PDF 

FabO allowed learners to laser-cut and 3D print objects from their favorite games, such as collectibles or custom controllers crafted from game-related elements. A key novelty of this work is using computer vision to allow seamless integration of fabrication moments within existing games without needing access to the game’s source files.

(published in ACM Creativity and Cognition'22)

 DOI | PDF | Project | Talk  | MIT News

This work employs Papert’s theory of constructionism to develop a game-based framework for learning fabrication and implements it in a toolkit to integrate fabrication activities within existing video games to teach digital fabrication. 

(published at ACM CHI ’22)

 DOI | PDF | Project | Talk 

While much of the design and engineering of current learning systems is driven by the advances in technology, we can reimagine these systems by reorienting the design goals around constructivist and sociocultural theories of learning to support learning progression, engagement across artistic disciplines, and designing for inclusivity and accessibility.  

(led the workshop at ACM CHI'22)

 DOI | PDF | Workshop Website

SensorViz is a visualization tool that supports novice makers during different prototyping stages with sensors. SensorViz provides 3 visualization modes: datasheet specifications, sensor interaction with the environment via AR before building the physical prototype, and live/recorded sensor data to test the assembled prototype. 

(published in ACM DIS'22)

 DOI | PDF | Project