Toggle navigation menu
Contact Us
Human ↔ Machine Interface

Energy-Harvesting Intelligence

Ambient energy models and firmware policy so sensing nodes can run when a primary battery is the bottleneck.

Energy-Harvesting Intelligence
RESEARCH THESIS

Power policy for constrained nodes—not the sensor and not the clinical loop.

Wearables, implants, and long-life sensors often fail for a simple reason: energy. Charging, swapping cells, and thermal limits constrain what can be sensed and how often intelligence can wake up.

This program studies intermittent computing, hybrid harvesters, and power management that make useful work possible from RF, piezo, thermal, and kinetic sources. The goal is intelligence that stays available without a primary battery as the only answer.

Findings here feed aFlex-class hardware, WearIQ, and any product that must live on the body or in the field for years.

Depth here is energy accounting: cold-start voltage, storage ESR, conversion efficiency across light/vibration/thermal regimes, and the policy that decides when sensing is allowed. Most of that work is simulation and logged analytics; partner labs supply transducer and IC characterisation when we need device-level confirmation.

Scope: This program owns joules: harvesting, storage, and duty-cycle policy. What is sensed at the tissue edge sits in In-Body Intelligence; what decisions those samples drive sits in Closed-Loop Human Intelligence.

WHAT WE BUILD TOWARD

Durable layers of the program.

Energy layers that stay useful across many sensing products—without absorbing biosignal analytics or implant packaging.

  • RF harvesting
  • Piezoelectric harvesting
  • Thermal harvesting
  • Kinetic / motion harvesting
  • Intermittent computing
  • Ultra-low-power MCUs
  • Energy-aware scheduling
  • Hybrid storage
  • Power-management ICs
  • Battery-free sensing nodes
METHODS & ALGORITHMS

Ambient power is a control problem, not just a transducer.

Energy-harvesting intelligence couples transducer models with firmware policy. We co-design MPPT, storage health, and duty cycles in simulation first, then validate critical points with partner evaluation kits or health-device benches.

We treat harvesting, storage, and the application as one closed energy loop: every sample, radio wake, and crypto operation is priced in joules.

Algorithms we study and implement

  • Perturb-and-observe and fractional-open-circuit-voltage style MPPT for PV and similar sources
  • Piezoelectric and TEG equivalent-circuit models under intermittent excitation
  • Finite-state energy managers: cold-start → charge → sense → transmit → sleep
  • Adaptive duty cycling driven by storage voltage and predicted arrival of energy
  • Supercapacitor / hybrid storage charge models with leakage and ESR awareness
  • Event-triggered sensing so scarce energy buys the highest-value measurements

Things we measure

  • Cold-start success under declared irradiance or vibration profiles (sim + partner kit logs)
  • Conversion efficiency across operating points—not peak datasheet claims
  • Autonomy under indoor / on-body energy traces from models or partner captures
  • Brown-out and recovery behaviour in firmware-in-the-loop tests
  • Energy cost per sample and per uplink packet from instrumented simulations
HOW WE EVALUATE

What counts as proof in an energy-limited system.

We prefer boring, plotted evidence: voltage vs. time under a named ambient profile—from simulation or partner kits—with the sensing schedule declared up front.

  1. Published ambient profile (lux, vibration, ΔT) used for every autonomy claim
  2. Measured cold-start and blackout recovery—not only steady-state efficiency
  3. Side-by-side comparison of fixed vs. adaptive duty cycles on the same hardware
  4. Failure modes listed: shade, misalignment, temperature swing, storage ageing
  5. A product bridge only when energy budgets fit a WearIQ or medical-edge form factor
In-Body Intelligence
Tissue interfaces need power strategies that stay tiny and safe.
Closed-Loop Human Intelligence
Closed loops only work if the energy budget can sustain them.
WHERE THIS SITS

An open field, not a closed stack.

Ambient-power research spans semiconductors, MEMS, and medical wearables. We partner across those layers rather than reinventing them alone.

Adjacent work

  • Energy-harvesting chip vendors
  • Ultra-low-power MCU makers
  • Academic intermittent-computing groups
  • Battery-free IoT startups
  • Medical wearable power research

Partners and complements

  • Semiconductor partners
  • MEMS manufacturers
  • Wearable OEMs
  • aFlex Ready Hardware
  • WearIQ
  • AL DeviceHub
  • Connected Care
HOW IT CAN GROW

A path that compounds over time.

From scavenged microwatts to reliable sensing windows—and eventually to on-device decisions that still respect energy reality.

  1. Reliable harvest + storage for short sensing bursts
  2. Duty-cycled intelligence that wakes with the right signal
  3. Hybrid harvesters tuned for wearable and implant-adjacent use
  4. Productised modules for aFlex and WearIQ platforms
  5. Field validation across multi-year deployments
NEXT STEP

Explore a collaboration.

Tell us about the sensing problem, robot behaviour, or long-horizon direction you want to shape with us.