Alexandra W.D. Bremers, Ph.D.
I make AI & Robotic systems work with people in the real world.
Researcher, designer, and technologist bridging human-computer interaction, robotics, and creative systems. PhD from Cornell plus 3 years industry R&D experience.
About
My PhD dissertation at Cornell, Towards Mixed-Initiative Machines for Creative Tasks, investigates how interactive and intelligent machines can support human workflows during physical tasks, especially fabrication. I integrate human-computer interaction, design, and systems development, working with collaborators across academia and industry.
Before Cornell, I spent three years as an HMI Researcher at Jaguar Land Rover, collaborating with Cambridge University on AR head-up displays and driver workload modeling. During my PhD, I interned at Walt Disney Imagineering (creativity support tools), Accenture Labs (task guidance systems), and Toyota Research Institute (computer vision).
My background spans Industrial Design (B.S., TU Eindhoven), Artificial Intelligence (M.S., Utrecht), and Information Science (Ph.D., Cornell), with a CS minor in computer graphics. I'm a Cornell PiTech AI in Arts & Culture Fellow.
Featured Work
Proactive Sewing Assistance (CoSew-4)
A Wizard-of-Oz study comparing proactive and reactive machine assistance during a sewing task with 20 novice participants. Proactive assistance, where the machine anticipated confusion based on behavioral cues, was perceived as significantly more timely without increasing annoyance. Resulted in the CoSew-4 synchronized multi-view video dataset.
Ph.D. Dissertation, Chapter 5
Instrumenting Machines for Creative Collaboration
How should fabrication machines be instrumented to capture human-machine interactions and support creative workflows? Multi-camera video analysis of novice users with tabletop machines. Includes the FabriCam-5 multiview video dataset.
Under review, 2025
Can Machines Tell What People Want?
The 'Hero' system for eliciting interactions around creative drawing machines. Explores how AI should gauge its actions by watching the user's response, since there is no "correct" answer in creative work.
HTTF 2024 · Paper
The BAD Dataset & BADNet
A dataset and model for detecting robot task failures through bystander facial expressions. Developed a framework for error recognition in HRI, collected the Bystander Affect Detection dataset, and trained BADNet to detect failures from facial reactions.
IEEE IROS 2023 · Paper · Project site
Collaborative Creative Machines
Envisioning and prototyping new ways of collaborating with creative machines, balancing automation and craftsmanship in mixed-initiative drawing systems. Built Wizard-of-Oz systems with varying degrees of autonomy.
HRI 2024 Workshop · Paper
All Projects
Proactive Sewing Assistance (CoSew-4)
Wizard-of-Oz study comparing proactive vs. reactive machine help during sewing. Proactive assistance was perceived as significantly more timely without increasing annoyance.
Instrumenting Machines for Creative Collaboration
Multi-camera video analysis of novice users with tabletop fabrication machines.
Can Machines Tell What People Want?
The 'Hero' system: AI that gauges creative collaboration quality by watching user response.
Collaborative Creative Machines
Balancing automation and craftsmanship in mixed-initiative drawing systems.
Designing Expressive Machine Movement
Communicative properties of drawing machines, collaborating with dancers and comedians.
Tic-Tac-Toe Drawing Robot
A pen plotter plays Tic-Tac-Toe against humans using computer vision, with expressive movements that communicate confidence, hesitation, and frustration.
Understanding Maker Entrepreneurs
Interviews with 20+ maker entrepreneurs on the tensions between wanting machine support and preserving the satisfactions of skilled making.
Amada Factory Observation
Observational site visit at a sheet metal bending workshop, studying how operator expertise shapes desire for machine assistance across analog to digital interfaces.
Drawing with Machines
Artistic explorations with the AxiDraw pen plotter: watercolor, 3D drawings, co-drawing.
The BAD Dataset & BADNet
Detecting robot task failures through bystander facial expressions.
Anticipatory Reactions in HRI
Literature Review: Predicting robot task outcomes from instrumented bystander reactions.
Social Discomfort in Cars
Analyzing family car trips and designing conversational interventions for shared driving.
XR-OOM: MR Driving Simulator
Low-cost mixed reality driving simulator with virtual objects in real-world driving.
Modeling Driver Interruptibility
Bayesian workload estimation from vehicular signals at Jaguar Land Rover + Cambridge.
Perception of Perspective in AR HUDs
Depth perception experiments for augmented reality head-up displays. Covered by Forbes.
Situated Audio Task Guidance
Wizard-of-Oz system testing audio cues for physical assembly guidance, with Accenture Labs.
MakerLab @ Cornell Tech
Supermaker volunteer operating the lab and documenting end-to-end Form 4 resin printing workflows for custom jewelry prototyping.
Computation for Content Creation
MediaPipe face tracking with affine image transformations to dynamically map flower images to video, inspired by Arcimboldo. Cornell CS6682.
Tilt Shift NYC
Photoshop tilt-shift camera effect on a photo from the New York Times Building, using neural filters, generative fill, blur, stamp, and masking.
Street Fashion Analysis
KNN clustering-based color detection method for clothing in street-level image datasets, exploring how fashion trends move over space and time. With Toyota Research Institute.
How to Embroider a Radio
AM crystal radio receivers from machine-embroidered conductive yarn coils. B.Sc. thesis.
Chiaroscuro: A Light Installation
Retro-reflective installation: light as a private experience in a shared space.
Meat the Future: Speculative Design
Fictional cookbook questioning lab-grown meat. Exhibited at Dutch Design Week.
Energy Belt Speculative Prototype
A belt that charges your phone on belly fat. Neopixel LEDs + Arduino.
Smart Home Interface Design
GUI as design probe for qualitative interviews at NTU IoX Center, Taipei.
CoffeeChat: Design for Debate
Speculative app questioning computer-mediated communication.
Wandi & Kepi: Superhero Sound Effects
Wearable cape + wand with superhero sounds for children with disabilities.
Fast, Intense, Repulsive
Expressing aesthetic qualities through clay, wood, and carved potato with alpaca wool.
A Postmodern Entrance
Conceptual entrance for the Groninger Museum using cactus-as-column metaphors.
Research Themes
Creative Machines
Mixed-initiative AI systems that collaborate with people on creative tasks like drawing, fabrication, and design.
Sensing & HRI
Detecting human intent and recognizing errors through social cues, facial expressions, and multimodal sensing.
Automotive HMI
In-vehicle interfaces, driver workload modeling, AR head-up displays, and mixed-reality driving simulation.
AI & Task Guidance
Situated conversational agents and audio interfaces that guide people through physical assembly tasks.
Physical & Speculative Design
Tangible prototypes, speculative objects, and installations exploring the intersection of technology and experience.
Experience
Education
Get in touch
I'm on the job market for R&D roles in AI and robotics, starting July 2026. NY/CA/MA preferred, open to relocation. Authorized to work in the US.