ExoLab.LLC — Engineering the Future of Human Mobility

Roadmap · 2026–2027

Five phases to walking often.

Getting Greg walking with a commercial exoskeleton by end of 2027, while building Layer 1/2 control intelligence for ataxic gait that doesn't yet exist. The personal test case drives the research.

Evaluate In progress · May–Aug 2026

Map the landscape before committing money or direction. Understand what's available, what's missing, and where the biggest contribution gap sits.

Inventory commercial exoskeletons — MMI access, programmability, SCA3 fit
Explore funding: Medicare/insurance (currently SCI-only), NIH SBIR/STTR, NSF, DARPA, National Ataxia Foundation
Determine whether Layer 1 intent detection is solved for ataxic gait (answer: it is not)
Establish community presence — IROS 2026 (Pittsburgh), OpenExo GitHub, NAF network, Shirley Ryan AbilityLab
Decide role: developer, investor, or hybrid (leaning: hybrid, primary developer)

Device	Maker	Intent Interface	SCA3 Fit
ReWalk 7	Lifeward	Wrist control + tilt sensor; cloud-connected; stairs/curbs	Requires reliable lean mechanics
Indego Personal	Ekso Bionics	Center-of-pressure at hip; modular; no stairs	Moderate — passive initiation helps
Atalante X	Wandercraft	Fully autonomous balance — hands-free, no crutches; AI-driven gait generation	Strong candidate — removes crutch dependency, critical for ataxia
Eve	Wandercraft	Self-balancing personal exo — NVIDIA Jetson AI, no crutches, 5-session onboarding; home & community use	Highest relevance — personal/home device targeting neurological gait, ~2026 FDA filing
HAL	Cyberdyne	Surface EMG — reads bioelectric intent signals from skin	Best theoretical fit for ataxic gait
OpenExo	NAU / U Mich (OSS)	Fully programmable (Teensy 4.1 + Arduino Nano 33 BLE)	R&D platform — Layer 1/2 dev target

Answer & Plan Aug–Dec 2026

Convert Phase I questions into decisions. Secure clinical access. Define the R&D scope. Build the funded plan.

Clinical evaluation at certified exoskeleton center (Sheltering Arms, Richmond VA or JHU affiliate) — establishes candidacy and research protocol access
Device access strategy: clinical research protocol (preferred) vs. personal purchase (~$100–200K)
Grant proposal(s) submitted — SCA3 + ataxic gait control is an underserved intersection
ExoLab Layer 1/2 scope defined: gait parameters to instrument, control architecture to prototype
Development environment live: OpenExo fork, Teensy board, Azure Kinect for gait capture

Note: Current FDA approvals cover SCI, not cerebellar ataxia. Positioning this as research — not just personal use — is the fastest path to device access while generating publishable data.

III

Implementation Q1–Q2 2027

Acquire the device. Begin instrumented walking. Get the control prototype running.

Exoskeleton acquired; certified companion trained; first assisted walk completed
Home environment on Canby Road adapted for device use
OpenExo dev environment operational in Crostini/Linux
Initial gait capture sessions using sensor suite; SCA3 dataset begun
Layer 1/2 control prototype running in simulation

Iteration Q2–Q4 2027

The engineering loop. Walk more. Walk better. Build smarter.

Instrument → Walk → Capture data → Refine control parameters → Repeat
Increasing frequency of walking sessions; improving gait quality; reducing fall risk
First findings published — even informally via OpenExo community
Layer 1/2 prototype tested on actual device; clinical collaborators engaged

Practice, Walk & Dance Dec 2027 →

Walking is routine. The R&D has produced something shareable. The mission clock hits zero.

The goal isn't to walk once. It's to walk to the mailbox. Walk at a friend's gathering. Dance at a grandkid's birthday. Walk because it's Tuesday.

Walking is a regular part of life — not a clinical event
ExoLab has a publishable control system contribution in the SCA3/exoskeleton intersection
Open-source contribution to OpenExo; established position in the field

// Hardware Context

Exoskeleton Market Landscape

ExoLab's Layer 1/2 control systems are designed to run on top of commercial exoskeleton hardware — not replace it. The table below surveys the current market: consumer hip-assist devices and FDA-cleared medical exoskeletons.

Consumer / Fitness

Medical / Clinical

Company / Model	Price (USD)	Motor Count	Peak Power	Max Torque	Weight	Battery Range	Modes / AI	App / Interface	Use Case	Notes
▸ Hypershell — Consumer Fitness & Outdoor
Hypershell X Go Consumer	$899	2 hip motors	400 W	~16 N·m	~2.2 kg	15 km	6 modes, AI MotionEngine LIT	Hypershell+ app	Entry walking/hiking	IP54, -10°C, foldable. No charger included.
Hypershell X Pro Consumer	$899	2 hip motors	800 W	32 N·m	~2.2 kg	17.5 km	10 modes (Walk, Run, Stairs, Hills…), AI MotionEngine	Hypershell+ app	Hiking, cycling, mixed terrain	IP54, -20°C. 2 ms response. Carbon fiber + aluminum frame.
Hypershell X Carbon Consumer	$1,299–1,599	2 hip motors	800 W	32 N·m	1.8 kg	17.5 km	10 modes, AI MotionEngine	Hypershell+ app	Hiking, extreme terrain	Carbon fiber + titanium alloy. 4,000 km durability rating.
Hypershell X Ultra Consumer	$1,799–1,999	2 hip motors	1,000 W	~22 N·m	~2.3 kg	30 km	12 modes, AI MotionEngine	Hypershell+ app	Long-distance hiking, daily use	IP54, -20°C. Titanium + carbon fiber. 12 sensors incl. barometer, gyro.
Hypershell X Pro S NEW 2026 Consumer	$999	2 hip motors	800 W	18 N·m	~2.1 kg	17.5 km	HyperIntuition™ AI (0.31s sync), TÜV Rheinland verified	Hypershell+ app	Lighter outdoor activity	Launched May 2026. Soft package. Max 20 km/h.
Hypershell X Max S NEW 2026 Consumer	$1,499	2 hip motors	1,000 W	22 N·m	~2.2 kg	30 km	HyperIntuition™ AI, 36× processing leap	Hypershell+ app	Heavy outdoor / load carry	Titanium alloy waist/back. SpiralTwill 3000 carbon fiber. Max 25 km/h.
Hypershell X Ultra S NEW 2026 Consumer	$1,999	2 hip motors	1,000 W	22 N·m	~2.2 kg	30 km × 2 batteries	HyperIntuition™ AI, 0.31s intent sync	Hypershell+ app	Elite outdoor / search & rescue	3D-printed titanium hip tube. 0.7 mm wall carbon. ↓HR 42%, ↓O₂ 39%.
▸ DNSYS — Consumer Fitness & Mobility
DNSYS X1 Carbon Consumer	$999–1,099	1 hip motor	900 W (1.2 HP)	40 N·m	1.6 kg	20 km / 7 hr	AI intent prediction, adaptive learning, dual-core 240 MHz	DNSYS app (iOS & Android)	Hiking, running, daily mobility	3s battery swap. 9 safety modules. USB-C PD fast charge.
DNSYS X1 Carbon Pro Consumer	$1,899	1 hip motor	900 W (1.2 HP)	40 N·m	1.6 kg (actual ~2.2 kg)	40 km (2 batteries)	AI intent prediction, regenerative mode, 9 safety modules	DNSYS app	Heavy hiking, workout, running	Titanium alloy + carbon fiber. 3s battery swap. Max speed 27 km/h.
▸ AstroShell — Consumer Fitness
AstroShell Alpha 1 Consumer	$1,099	2 motors (est.)	1,000 W peak	40 N·m continuous	2.0 kg	24 km	AI Active System, adaptive assist	Companion app (est.)	Hiking, walking endurance	Aerospace magnesium alloy. Swappable "pocket" batteries.
▸ Ascentiz — Consumer Modular / Open-Source Platform
Ascentiz H1 Pro (Hip Module) Consumer Open-Source BodyOS	$1,049 Early backer $699 · Kickstarter #1 funded exo	1 hip motor (quasi-direct-drive)	900 W (1.2 HP)	36 N·m peak	1.75 kg (w/o battery)	~78 Wh battery	AI Motion Cortex, 10+ motion scenarios, 0.2s recognition	Companion app + open-source BodyOS SDK	Hiking, running, daily mobility	iF Design Award 2026. 35% effort reduction. Folds to A4. ⚡ BodyOS = ExoLab integration opportunity.
Ascentiz K1 Pro (Knee Module) Consumer Open-Source BodyOS	$1,149 Early backer $799	1 knee motor (cable-drive)	900 W (1.2 HP)	48 N·m peak	~2.25 kg (w/o battery)	~78 Wh battery	AI Motion Cortex, 10+ motion scenarios, 0.2s recognition	Companion app + open-source BodyOS SDK	Endurance, load-bearing, joint support	Cable-drive optimized for knee biomechanics. Supports up to 216 lbs load.
Ascentiz H+K Bundle Consumer Open-Source BodyOS	$1,498 Early backer $1,298	2 motors (1 hip + 1 knee)	900 W (1.2 HP)	36 N·m (H) / 48 N·m (K)	1.75–2.25 kg per module	~78 Wh per module	Full Ascentiz AI Motion Cortex; modules swap one at a time via Exo-Belt hub	Companion app + open-source BodyOS SDK	Full-activity coverage — switch hip/knee per terrain	World's first modular exo system. BodyOS enables custom control layer development.
▸ Lifeward (formerly ReWalk Robotics) — FDA-Cleared Medical
Lifeward ReWalk 7 Medical FDA Cleared	~$75,000–95,000 Medicare reimbursable; $94,617 CMS rate, 20% copay	4 motors (2 hip, 2 knee)	N/A	Not published	~23 kg (51 lbs)	Session-based	2 customizable walking speeds, cloud connectivity, smartwatch display	MyReWalk app, crutch control, smartwatch	SCI T7–L5, home & community ambulation, stairs & curbs	FDA cleared March 2025. First device with FDA stair clearance.
▸ Ekso Bionics — FDA-Cleared Medical
Ekso Bionics Ekso Indego Personal Medical FDA Cleared	~$75,000–95,000 Medicare, VA, workers comp coverage	4 motors (2 hip, 2 knee)	N/A	Limited (BLDC flat motors)	~14 kg (31 lbs)	Session-based	Wireless software control, individualized gait settings	iOS & Android wireless app	SCI T3–L5, home & community; NO stair clearance	Lightest medical exoskeleton. Modular 5-piece design. Carbon-fiber thermoplastic.
Ekso Bionics EksoNR Medical FDA Cleared	~$100,000+ Clinical/rehab centers only	4 motors (2 hip, 2 knee)	N/A	Not published	~20 kg (44 lbs)	Session-based	Multiple gait modes, stroke/SCI/ABI/MS modes	iOS app (therapist-controlled)	Rehab only: stroke, SCI, ABI, MS	Only exoskeleton FDA cleared for ABI and MS. Not for home use.
▸ Wandercraft — FDA-Cleared Medical & Personal (Hands-Free)
Wandercraft Atalante X Medical FDA Cleared	~$150,000–250,000 Clinical/rehab centers only; not for personal purchase	12 actuated joints (6 per leg: hip flex/ext, hip abduction, knee, ankle)	Not published	Not published	~50 kg (110 lbs)	Session-based	Fully autonomous dynamic balance — no crutches or parallel bars; AI-driven gait; therapist tablet control	Therapist tablet interface	Rehab: stroke, SCI, neurological gait impairment; NO home use	FDA 510(k) cleared Nov 2023. Only fully autonomous hands-free exo. Strong SCA3 candidate due to no crutch requirement.
Wandercraft Eve Medical Clinical Trial	TBD Personal/home use; Medicare coverage anticipated; ~2026 commercial launch	Not published	Not published	Not published	Not published	All-day personal use	Self-balancing AI gait (NVIDIA Jetson edge compute); no crutches; 5-session onboarding; real-time adaptive gait across surfaces	Personal user interface (details TBD)	Personal/home: SCI T6+; neurological gait impairment; community ambulation	In clinical trials (Kessler Institute, VA Medical Center NY). FDA filing targeted post-trial. Highest personal-use relevance for SCA3 — self-balancing removes crutch dependency entirely.

// Research Notes

ExoLab Findings

AI-assisted knowledge-base entries documenting device research, control-system analysis, and clinical evidence — generated in session and published here.

2026-06-14 · KNOWLEDGE-BASE ENTRY · DOMAIN: EXOSKELETONS FOR THE DISABLED · GENERATED WITH CLAUDE

Exoskeletons for Neuro-Challenged (Non-SCI) Adults — Expert Briefing

Core Reframe: Output Failure vs. Coordination Failure

The field splits on one question: is the deficit a failure of motor output or of motor coordination?

OUTPUT FAILURE — SCI · SEVERE STROKE · LATE MS

Motor command weak/absent. Robot's job = substitution — drive joints through pre-planned trajectory. Position control, fixed gait cycle, "puppet" model. This is what almost all famous exos do.

COORDINATION FAILURE — CEREBELLAR ATAXIA (SCA3, MSA-C) · PARTLY PARKINSON'S

Strength largely intact; timing/scaling/feedback broken. Ataxic gait = unstable walking, increased step width, balance deficit, high stride-to-stride variability. Robot's job = augmentation/stabilization — damp error, supply rhythm, constrain sway without overriding intent. "Dance partner" model.

A device transformative for a paraplegic can be useless/destabilizing for ataxia because the control philosophy targets a different failure mode.

Map of the Landscape

Rehab vs. personal: most devices are clinic gait-trainers. Of legacy devices, only ReWalk and Indego are FDA-approved for personal/home use — both indicated for SCI.
Substitutive vs. synchronization control: most rigid exos are substitutive. A small (mostly Japanese) line uses synchronization/CPG-based control — the correct architecture for ataxia, and the one that maps to ExoLab Layer-1 (intent) / Layer-2 (gait planning).

Device Comparison (Neuro, Non-SCI Focus)

DEVICE	CONTROL	FDA NEURO INDICATIONS	HOME USE	ATAXIA
EksoNR	Substitutive, adaptive assist	Brain injury, MS, stroke + SCI (first FDA-cleared for all)	Clinic only	LOW
Keeogo	Assistive knee power, follows user motion	510(k); MS, PD, ataxia listed	Portable/home	HIGH
curara ★	Synchronization / CPG	Not FDA-cleared (Japan)	Designed for home	HIGHEST
Wandercraft Eve	Self-balancing, hands-free	In trial (SCI/paraplegia), not marketed	Designed for home	UNKNOWN
HAL (Cyberdyne)	Bioelectric (EMG) intent	Post-stroke paresis, neuromuscular paraplegia, CP, spastic paraplegia	Clinic-centric	MODERATE

Keeogo — The Responder Window

Provides dynamic stability + minimally necessary assistance (marching, lunges, step-ups, walking, stairs). Critical finding: responders had moderate impairment — Berg Balance Scale 46–51 or Timed Up and Go 8–12 s. Too functional or too impaired = no benefit. Progressive disease eventually walks a person out of the window. This is the key number for any assistive (non-substitutive) device.

curara — The Device to Know for Cerebellar Disease

Assists hip + knee; supports rhythmic gait via synchronization control based on a central pattern generator (neural oscillator).
Detects interactive force from slight wearer movements to reflect intent; coordination tunable via synchronization gain.
Soft-ish architecture: actuators on molded joint supports, no metallic exoskeleton frame (rigid frames fight ataxic micro-corrections).
Evidence: 2018 study (n=12 SCD) improved walking smoothness; 2021 single-arm study (n=20 SCA/MSA-cerebellar, 15-day training). Korean ANGELLEGS case report (SCA) showed balance/gait gains persisting at 4-week follow-up.

ExoLab takeaway: existence proof that CPG-synchronization works as intent-respecting control for ataxic gait — the Layer-1 → Layer-2 decomposition. Open problem: generalizing from rhythmic/treadmill gait to free-living, variable-terrain, start-stop ambulation.

Evidence Reality Check

Strong RCT-grade evidence = stroke + SCI rehab gait training.
Cerebellar ataxia exo literature = case reports + single-arm studies (n=1 to ~20), no controlled trials, short durations. "Improved while wearing it" ≠ durable rehab effect (curara authors flag this explicitly).
Moderate evidence that rehabilitation in general helps cerebellar postural control — but that argues for intensive PT, not specifically robots.
Honest position: robotic gait training in ataxia is plausible/promising, not established. Choose by responder-window logic, not brand.

Access & Economics

Medicare pathway = brace benefit. CMS-1780-F classifies powered exos that support weak limbs as braces; lump-sum reimbursement under HCPCS K1007 (est. for ReWalk).
Framing built around weak limbs and SCI — poor fit for ataxia (limbs not weak).
Wandercraft Eve trials = SCI/paraplegia/tetraplegia, not coordination disorders.
Net: as personal exos arrive, both coverage logic and trials point away from the neuro-coordination population. This is the exact gap ExoLab and the Wandercraft discovery call are probing.

Threads to Pull Next

Full curara / AssistMotion technical lineage; whether type-4+ units are obtainable outside Japan.
Whether Eve's self-balancing layer could be repurposed as an ataxia balance aid despite paralysis-oriented marketing (open, ExoLab-relevant).

Sources

Wearable Robots for Rehab and Assistance of Gait — PMC12411867
Robotic Exoskeleton-Assisted Gait Training in SCA (ANGELLEGS) — PMC8309925
Robot Assisted Gait Training in a Patient with Ataxia — PMC9326713
curara / synchronization control in SCD — J NeuroEng Rehabil 2018 (PMC6146529); BioMed Eng OnLine 2021 (PMC8424896)
Keeogo responder window — PhysioFunction / McLeod et al.; Nature Sci Rep 2024 (Parkinson's)
FDA clearances — Ekso Bionics; Molina/Premera/FEP medical policies; Cyberdyne HAL K233695
Wandercraft Eve — Series D coverage 2025; ClinicalTrials NCT06715631, NCT06777576
Medicare brace benefit — CMS-1780-F; Noridian K1007; Lifeward/ReWalk releases

Engineering theFuture of Walking

A DSP Engineer Who Needs an Exoskeleton