De-Abstracting the Machine: Manav Goel on Tabletop Manipulation, Open Hardware, and the Physical AI Frontier

01The Pedagogical Bottleneck: Moving Beyond the Chatbot

Artificial intelligence education has spent years locked behind glass screens. Students are taught to prompt language models and tune digital algorithms, yet the messy, dynamic laws of physical reality — gravity, inertia, gear backlash, and friction — remain entirely abstract. When the RoboRoots Robot and Physical AI Hackathon convened at the African American Art & Culture Complex (AAACC) in San Francisco, the central engineering challenge was clear: how do you introduce middle and high school students to the complete end-to-end Physical AI workflow in a single day?

The answer did not lie in multi-million-dollar proprietary setups. It began with an accessible bridge: showing students that physical robotics starts with open-source 3D-printable components, affordable smart servos, and edge microcontrollers (like ESP32s and Seeed edge hubs), giving them the foundation to eventually design, modify, and actuate their own custom robotic mechanisms from scratch.

02Master the Arm, Command the Humanoid

A tabletop robotic manipulator is not just a desk toy; it is the fundamental building block of embodied intelligence. The exact mathematical principles governing a 6-Degree-of-Freedom (6-DoF) arm — forward and inverse kinematics, joint coordinate limits across X, Y, and Z frames, torque control, and end-effector grasping — are identical to the spatial dynamics that drive 30-DoF bipedal humanoids (like the Unitree G1) and agile quadrupeds (like the Go2). By understanding how a robotic arm reaches, holds, and stabilizes, youth builders unlock the universal mechanics behind physical machines navigating complex human spaces.

03Stellar Mentorship: Manav Goel at the Arena Bench

Transforming dense kinematics into intuitive understanding requires mentors who have walked the path. Leading the physical arm track at RoboRoots was Manav Goel, an AI engineer with Seeed Studio. Manav guided cohorts of youth builders across various manipulator setups — from the accessible, open-source Seeed Studio SO-101 leader-follower arm pair to high-precision industrial robotic arms. Rather than lecturing on abstract formulas, Manav walked students through the complete live pipeline:

1. STEP 01

   Calibrate

   SO-101 leader-follower motor ports

2. STEP 02

   Demonstrate

   Grasp foam cubes & erasers via teleop

3. STEP 03

   Record

   Synchronized 1080p multi-camera LeRobot datasets

4. STEP 04

   Train

   Push demonstrations through SenseCraft

5. STEP 05

   Deploy

   Autonomous physical inference on the arm

04The Builder's Journey: Choosing Physical Over Agentic

During his on-site interview with the RoboRoots broadcast desk, Manav shared what originally drove him into the field. Pursuing his Master's in Artificial Intelligence at San Jose State University (SJSU), he found himself disenchanted with software-only agents:

When asked about the future of physical AI, Manav emphasized the shift toward automating mundane, repetitive, and hazardous labor:

05Sparking Youth Imagination: Questioning the Interaction Loop

The ultimate goal of RoboRoots was never to teach students to copy an existing script — it was to spark community innovation. As student builders completed their teleoperation sessions, Manav and the mentoring team posed critical questions:

• >Why do we teleoperate arms using a passive leader rig?
• >Could we capture demonstrations with spatial VR headsets, wearable sensor gloves, or natural computer-vision hand-tracking?
• >How would you redesign the end-effector to harvest delicate crops, sort recyclable municipal waste, or assist elders with mobility?
• >What changes when this arm is mounted to a mobile rover for urban agriculture or senior assistive care?

By challenging students to critique the tools they were using, the hackathon transformed them from operators into system architects capable of inventing new interaction paradigms for their neighborhoods.

06The Launchpad: Deep Gratitude to Seeed Studio

RoboRoots extends our profound gratitude to Seeed Studio and Manav Goel for their dedication and hardware leadership on the tournament floor. Seeed Studio's robotic and edge AI ecosystem — especially their SenseCraft AI workflow platform and open-hardware manipulators — stands out as the premier launchpad for education, research, and community-driven physical AI. By demystifying the barrier between software and hardware, Seeed Studio is helping us cultivate an ecosystem where every young builder can turn imagination into physical reality.