2009 Tulip
TU Delft Robocup humanoid soccer · with Prof. Pieter Jonker
Early predecessor of modern humanoid programs — bipedal control, perception-action loops, team play.
Dr. Ir. Aswin Chandarr
Bringing Physical AI to the real world.
Eighteen years architecting robots that ship — humanoids, autonomous shuttles, hospital systems, industrial arms in tunnels. Today I lead VLA-driven social understanding, trust-building, and dynamic behavior on a research humanoid and its market-ready desktop sibling. PhD-grade depth in embodied AI, MBA-grade discipline on P&L. Author of Inevitable AI.
01 — Eighteen years of robots
Every robot I've developed, architected and delivered.
2009–2025 · humanoids, autonomous shuttles, hospital systems, industrial arms in tunnels — plus a book. The visual inventory, in chronological order.
- 2009
-
2012 PhD — Cognitive Robotics
TU Delft · Jonker / Rudinac
Doctoral work on perception, learning, and human-aware behavior on real platforms.
-
2014 LEA
Co-founder, Robot Care Systems (NL)
Elderly walking & companion robot. CE / medical-device track.
-
2016 WePod
RRC Robotics / TU Delft · Tech architecture & multi-sensor fusion
Public-road autonomous shuttles. Among the first to deploy NVIDIA DrivePX in production.
-
2009 All Terrain Robot DFKI
Researcher
Simultaneous Wheeled Legged Robot (SWLR) for outdoor inspection and surveillance. Motion Planning
-
2018 Pet AI Companion
Technical Business Manager ·Jonker Makis Robotics (NL)
Indoor mobile security robot for warehouses and large office spaces.
-
2019 SAM-UVC
Founding CTO, Loop Robots (NL)
Autonomous UVC hospital disinfection. 3 published patents on architecture, dosage mapping, mixed-mode autonomy.
-
2023 The Inevitable AI
Author · published via GAITI
A first-principles guide to AI for the people who have to decide — professionals, enterprises, investors, founders, policy makers.
-
2024 Hubertustunnel
Advisory · Grimbergen Industrial Systems · Den Haag
World-first adaptive milling robot. 26,000 m² resurfaced at ±5 mm with real-time scan-based path planning.
-
2025 RIA
Head of Engineering & Product · Machani Robotics
Research humanoid in the Sophia lineage. VLA for dynamic HRI, social understanding, trust building.
-
2025 CC — Cognitive Companion
Head of Engineering & Product · Machani Robotics
Desktop product that ports RIA's social abilities into a market-ready form. 0→market in 6 months.
Drag, scroll, or swipe →
01 — Thesis
Physical AI is the next decade's most consequential shift — and it isn't going to ship itself. Most labs write papers. Most operators scale software. Very few have spent eighteen years where the demos meet the loading dock: failure modes, certification, 2 a.m. field calls, the gap between a benchmark and a production line.
My career spans two eras of the field. I built robots on the classical stack — SLAM, multi-sensor fusion, motion planning, regulatory certification — for the first decade, and shipped them. I'm building them on the Physical AI stack — VLA, embodied learning, synthetic-data strategy, foundation-model fine-tuning — now. Most people are fluent in one. The work ahead demands fluency in both, and the bridge between them is where I work.
At Machani Robotics, the current frontier is VLA for social understanding (what does the human in front of the robot actually want), trust building (does the robot earn the right to be in the room over time), and dynamic behavior (does it respond to a moving person and a moving world, not a fixed script). These are the parts of embodied AI that benchmarks don't measure and demos don't show.
I've co-architected an adaptive milling robot that resurfaced 26,000 m² of safety coating inside a Dutch highway tunnel on millimeter precision. Integrated NVIDIA DrivePX into shuttles that carry real passengers between Schiphol and its parking lots. Co-founded a care robot that walks with the elderly in their living rooms, founded a disinfection company that protects hospital wards from infection, and now lead CC, a desktop cognitive companion for older adults — taken from concept to market in six months at Machani Robotics.
Different robots, one job. Building machines that work the day after the demo.
02 — Shipped in the wild
Three pillars. One job: make Physical AI work outside the lab.
Care robots, autonomous mobility, industrial & infrastructure. Different domains, same hard problem — reliability, safety, and human interaction under conditions that can't be faked.
Pillar 01
Care & human-centric robotics
LEA — elderly walking & companion robot
Co-founder, Robot Care Systems · The Netherlands · 2014–2019
Co-founded RCS to take LEA from a TU Delft research prototype into a CE / medical-device-track product for older adults living independently. LEA physically supports walking, reminds, calls for help, and keeps a presence in the home. Built the technical organization, led integration of safety, UX, motor control, and regulatory pathway.
SAM-UVC — autonomous hospital disinfection
Founding CTO, Loop Robots · The Netherlands · 2019–2023
Founded Loop Robots and shipped SAM, an autonomous mobile UVC robot deployed in hospitals to map dosage and disinfect wards between procedures. Took the product through clinical pilots, dosimetry validation, and modular-accessory design for varied hospital floor plans.
- US20250319221A1 · UVC site disinfection
- US20240252703A1 · Modular service-robot architecture
- WO2024105448A8 · Mixed mode (auto + manual)
Pillar 02
Autonomous mobility in the real world
WePod — autonomous public-road shuttles
RRC Robotics / TU Delft · The Netherlands · ~2016–2018
Technical architecture, hardware/software integration, and multi-sensor fusion (camera, radar, LiDAR) for the first generation of WePod fixed-route autonomous shuttles. Operated routes between Schiphol Airport and its parking lots, the Wageningen University campus, and the Rotterdam metro–airport corridor — carrying real passengers under public-road conditions.
Among the first developers to deploy NVIDIA DrivePX in production, with direct technical feedback into NVIDIA's autonomy stack. Owned safety case, technical product management, and field-ops integration.
Schiphol autonomous tugbot
RRC Robotics · Research & feasibility
Feasibility and prototype work for an autonomous ground tug for airline ramp operations at Schiphol — moving baggage trains and equipment without a human driver, around active aircraft.
Pillar 03
Industrial & infrastructure
Hubertustunnel — adaptive milling robot
Advisory, Grimbergen Industrial Systems · Den Haag · 2024–2025
Co-architected an adaptive 6/7-axis milling robot on tracks for Vlasman B.V. / Strukton, contracted by the City of The Hague. Real-time scan-based path planning across the entire tunnel surface — no pre-programmed paths. The robot resurfaced the safety coating across 26,000 m² / 3,120 panels / 1,800 m of tunnel at ±5 mm precision. A world-first ("primeur") for tunnel renovation.
03 — Now
Machani Robotics — RIA & CC.
As Head of Engineering & Product at Machani Robotics (Bengaluru), I lead two paired platforms: RIA, a research humanoid in the lineage of Hanson Robotics' Sophia, and CC — the Cognitive Companion, a desktop product that ports RIA's social and cognitive abilities into a compact, deployable form factor. CC went from concept to market in six months.
CC — Cognitive Companion
CC is the deployable form of the same intelligence — a desktop companion for older adults that handles conversation, reminders, gentle escalation, and a quiet, durable presence in the home. Soft launch is live at cc.sonacares.com.
RIA — research humanoid
RIA continues the Sophia lineage into the VLA era. Where Sophia drove public discourse about social robotics on a scripted stack, RIA does the work behind that conversation: dynamic human–robot interaction grounded in a Vision-Language-Action policy. It is our research platform for the harder questions in embodied social intelligence — and the proving ground for everything we then port into CC.
The VLA work, in three lines
Social understanding
Inferring what the human in front of the robot actually wants — affect, intent, attention, conversational state — and grounding it in the physical context of the room. Models a person, not a prompt.
Trust building
Trust as a longitudinal property, not a single interaction. The robot earns the right to be in the room over days and weeks — predictable behavior, calibrated honesty about what it knows, graceful failure, durable memory of the person it serves.
Dynamic behavior
Acting fluidly in a moving world: a person changes posture, looks away, walks behind, leaves the room and returns. The behavior policy is closed-loop on perception, not a scripted dialogue tree — the difference between a demo and a deployment.
Patents filed at Machani
Three patents filed; under embargo.
Filed alongside the work, not after. Tracks the same architecture that ships in CC.
- Embodied AI — VLA architecture for socially-grounded action
- Deep personalization — long-horizon memory & preference modeling
- Proactive care — initiative, escalation, and safe handoff
05 — The book
The Inevitable AI
Art of Growth with Generative Intelligence
A first-principles guide to AI — where it came from, what it can and can't do, where the real opportunities and dangers sit. Written for the people who actually have to decide: professionals, enterprises, investors, founders, policy makers.
Most AI books are written either for engineers or for executives. This one was written for the rest of us — the people who have to live with the consequences.
Full sales treatment on /book — cover, ToC, blurbs, audiobook link.
06 — Writing, speaking & media
The gap between AI research and the systems people actually live with.
I write and speak for technical and non-technical audiences when I have something I haven't seen said. Topics I'm currently working on:
The Physical AI stack
What changes when robots learn the way LLMs do — and what doesn't.
Research-to-product
How to take embodied-AI research out of the lab without losing the science.
The economics of care robots
Why elderly-care and hospital robotics are the first real consumer markets for embodied AI.
Foundations of the Inevitable AI
The argument from the book, adapted for boards, investors, and operating teams.
07 — Advisory & investing
A small number of engagements where the work is technically hard and the stakes are real.
Physical AI startups, robotics teams inside larger companies, and funds underwriting the category. I'm useful when the question is one of:
Architecture & roadmap
VLA, embodied learning, classical-stack integration, sensor strategy, the build/buy line.
Research-to-product translation
Turning a lab demo into something a customer renews.
Regulatory & deployment
Medical-device pathway, public-road autonomy, industrial safety.
Technical due diligence
For funds evaluating robotics, embodied AI, or applied-research bets.
I currently advise on selected Physical AI programs in healthcare robotics, autonomous mobility, and industrial automation. I do not take engagements that conflict with my work at Machani Robotics, and I disclose every active role.
08 — Contact
Best way to reach me is email. I read every message; I reply to most.
- Email mail@aswinchandarr.com
- LinkedIn /in/aswinchandarr
- Google Scholar Scholar profile
For executive search, board, or investor outreach: send the role, the company, and the timeline in the first email. I'll come back within five business days.