AI in Robotics:
From
Perception to Autonomy
Practical training built around the real engineering decisions that go into making robots think. Covers sensor fusion, model deployment, and decision systems used in production environments today.
What the program covers
Eight weeks of structured sessions, each focused on a specific layer of an AI-powered robotic system. No filler — every module connects directly to the next.
Perception & Sensor Fusion
Working with LiDAR, depth cameras, and IMUs together. You'll implement Kalman filters and understand when sensor disagreement should trigger a safety halt rather than an average.
On-Device ML Deployment
Quantizing and running inference on edge hardware — NVIDIA Jetson, Raspberry Pi 5, and similar boards used in real field robots.
Motion Planning with Learned Models
Combining classical planners like RRT* with neural policies. When each approach fails and how hybrid systems handle the gap.
Safety & Failure Analysis
Designing systems that degrade gracefully. Includes fault injection exercises and review of real incident reports from autonomous vehicle deployments.
Program structure and pace
Each week runs two live sessions of 90 minutes each, held on fixed days so you can plan around them. Sessions are recorded, but the Q&A portions are live-only — the discussion that happens there tends to be where the most specific and useful answers come out.
Prerequisites are honest: you need working Python knowledge and some familiarity with linear algebra. The program doesn't re-teach those foundations — it uses them from the first session onward.
Duration
8 weeks · 16 live sessions
Group size
Maximum 18 participants per cohort
Format
Fully online · Live + async recordings
Certificate
Issued on completion of all modules
Yael Brandeis
Robotics Systems Engineer
12 years designing perception stacks for industrial and agricultural robots. Worked on autonomous harvesting systems deployed across three continents.
Yael leads the sensor fusion and safety modules. She brings real failure cases from field deployments — the kind of edge cases that don't appear in textbooks.
Tomás Ferreira
Machine Learning Researcher
Specializes in model compression and edge inference. Previously at a robotics lab in Porto, now consulting for hardware startups on ML pipeline design.
Tomás covers on-device deployment — the practical work of getting a model running on constrained hardware without losing the accuracy you trained for.
Noa Eldar
Computer Vision Specialist
Focused on real-time object detection and tracking for mobile robots. Built vision pipelines for logistics automation used in warehouse environments.
Noa handles the computer vision sessions — from camera calibration through detection and into the tricky problem of tracking objects across partial occlusions.