It’s Not Like The Movies: Managing Uncertainty When Tracking Objects in the Real World
Friday, April 17, 2026 · 1:45 PM – 2:15 PM · Auditorium 3
Hollywood makes tracking look effortless: a single object on a radar screen gives everyone perfect knowledge of the situation. Reality is messier. Applied math always comes with errors, and in the world of object tracking, issues like imperfect sensors, out-of-sequence measurements, and more make those errors compound. In this talk we’ll discuss how the Monte Carlo method can be essential for stress-testing your system. By running hundreds of randomized simulations, you can start to quantify error across different scenarios and move from the nagging feeling of “How wrong are we?” to a steady confidence in how well your system actually performs.
About the Speaker
Adrian Palacios
Data Science Manager at General Atomics Intelligence
Adrian Palacios is a Data Science Manager at General Atomics Intelligence, where he helps extend Python ecosystem tools for domain-specific problems like tracking filter optimization. He spent the previous decade at companies such as DigitalOcean, Twilio/SendGrid, and Zillow, leading analytics teams that transformed noisy, high-volume data into something stakeholders could actually act on. As a lifelong learner his interests span many things, from causal inference to the art of influencing without authority. Outside work, Adrian experiments with gluten-free baking and perfecting his taco recipes.