publications & presentations

2025

1. 

   Battery-free Computer Vision on Insect-scale Microrobots

   V. Arroyos* ,  M. Ibrahim* ,  E. Mensah* , and 3 more authors

   IROS (Late-Breaking Results), May 2025

   Abs

   The goal of this project is to enable the use of efficient on-device deep learning models for battery-free mobility-based sensing robots. We target the deployment of these models on MilliMobile, a one square centimeter microrobotic platform with only 512 KB of RAM and 1 MB of flash memory (millimobile.cs.washington.edu), making it challenging to run traditional models onboard. We address this challenge by integrating intermittent computing and motion with image-based sensing, ensuring that the system only performs high-power actions–like locomotion, inference, and communication–when specific conditions are satisfied. Event-based vision and intermittent computing approaches will also be leveraged to optimize the MCUs time in ultra-low-power modes. We train on insect images from Insect Detect - insect classification dataset v2 and achieve almost 70% accuracy after 500 epochs with milliWatts of power consumed when running the system on the nRF52840 microcontroller.

2024

1. Ethics

   Probing the Ethical Boundaries of Personalization: a Case Study of Twitter’s Recommendation Algorithm

   K. Feng ,  M. Ibrahim ,  and  J. Yoon

   Mar 2024

   Abs PDF

   Most online content platforms today are designed to maximize the time users spend engaging with their content. This engagement allows platforms to both serve advertisements and collect data from usage patterns to incorporate into their recommendation and per- sonalization algorithms. However, personalization algorithms are often opaque; surfacing relevant and interesting content to users at best, and constructing echo chambers in which users are not exposed to a diversity of opinions or beliefs at worst. This issue is exacerbated by the fact that many content platforms enable users to fine-tune personalization algorithms without much in the way of ethical guardrails, such as “liking” a post or selecting “see less/- more”. In this paper, we ask the question of whether there are ethical limits to personalization in content platforms. Twitter is a content platform with a wealth of publicly-available information surrounding its personalization algorithms, which we use as a case study for our investigation. We conduct a literature review of prior research in content personalization on social media and analyze published algorithmic personalization source code by Twitter along with related technical blog posts. We then identify ethically nu- anced components of Twitter’s content personalization pipeline and analyze them from the perspective of 6 ethical theories. We con- clude with a discussion of how developers can more deeply engage with ethical considerations when building personalized systems.

2. AI/ML

   TinyML for Efficient Detection, Classification, and Control

   M. Ibrahim

   Task Focused IR in the Era of Generative AI Workshop, Microsoft Research, Sep 2024

2023

1. Robotics

   Coin Copter: A Coin-Sized Helicopter

   M. Ibrahim ,  and  K. Hernandez

   Washington NASA Space Grant Consortium Symposium, Aug 2023