Module 4

Sensing & Information

Compound eyes, bat sonar, pit viper IR, lateral-line dipole detection, shark electroreception — biology's sensor arsenal and its engineering translations.

4.1 Compound Eyes: Insect Vision for Drones

Insect compound eyes consist of 100–30,000 independent ommatidia, each a small lens+photoreceptor unit sampling a narrow solid angle. Compared with a human single-lens eye, they have poor angular resolution (\(\sim 1^\circ\) vs 0.02 deg for humans) but enormous advantages: near-infinite depth of field, hemispherical field of view, extreme motion sensitivity (dragonfly tracking 300 Hz flicker), and no blind spots.

Derivation: Snyder's diffraction limit

Snyder (1979) derived the optimum ommatidium aperture \(d\) as a balance between diffraction (large \(d\) better) and inter-ommatidial angle\(\Delta\phi = d/R\) sampling (small \(d\) better). Setting the Airy-disk half-width \(\lambda / d\) equal to \(\Delta\phi\):

\( \boxed{d_{\text{opt}} = \sqrt{\lambda R}} \)

With \(\lambda = 550\) nm and eye radius \(R = 1\) mm:\(d_{\text{opt}} \approx 25\) μm — precisely matching bee anatomy. The honey bee has \(\sim 6000\) ommatidia in an eye of this radius, giving angular resolution \(\sim 1^\circ\).

Applications: CurvACE, panoptic cameras

• CurvACE (EPFL, 2013) — 630-pixel curved artificial compound eye, 180 deg FOV, on a 12.8 mm hemisphere.
• UIUC compound eye camera (Song et al. 2013, Nature) — photodetector array on a deformable substrate.
• Centeye Vision Chips — 1-D focal-plane optical-flow processors for insect-inspired MAV navigation.
• Cross-reference: Insect Biophysics for species-specific details.

Python

script.py80 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# Compound eye resolution limit (Snyder 1979)
# Interommatidial angle Delta_phi ~ sqrt(wavelength / R) for a curved eye
# Each ommatidium has aperture d -> diffraction limit

# For honeybee: ~6000 ommatidia per eye, eye radius ~ 1 mm, aperture 25 um
# Number of ommatidia N vs eye radius R under two constraints:
# 1. Diffraction-limited resolution
# 2. Minimum viable aperture

wavelength = 550e-9  # green light, m
R_eye = np.linspace(0.2e-3, 10e-3, 200)  # eye radius from 200 um to 1 cm

# Snyder's optimum ommatidium diameter
d_opt = np.sqrt(wavelength * R_eye) * 2   # m
# Angular resolution
delta_phi = d_opt / R_eye  # rad
# Number of ommatidia (solid angle of hemisphere / ommatidium solid angle)
# hemisphere = 2 pi sr, ommatidium = pi * delta_phi^2 / 4
N_ommatidia = 2 * np.pi / (np.pi * delta_phi**2 / 4)

fig, axes = plt.subplots(1, 2, figsize=(15, 6))
fig.patch.set_facecolor('#030712')

# Panel 1: angular resolution vs eye radius
ax1 = axes[0]
ax1.set_facecolor('#1a0a2e')
ax1.plot(R_eye*1000, np.rad2deg(delta_phi)*60, color='#e879f9', linewidth=2.5, label='Diffraction-limited (Snyder)')

# Real species (eye radius [mm], resolution [arcmin])
species = [
    ('Fruit fly', 0.3, 300, '#fbbf24'),
    ('Bee', 1.4, 60, '#f472b6'),
    ('Dragonfly', 2.5, 25, '#22d3ee'),
    ('Mantis shrimp', 3.0, 15, '#86efac'),
    ('Human eye', 11, 1.0, '#a78bfa'),  # single-lens, for comparison
]
for name, r, res, c in species:
    ax1.scatter(r, res, c=c, s=140, edgecolors='white', linewidth=0.8, zorder=5)
    ax1.annotate(name, (r, res), xytext=(r*1.2, res*1.2), color=c, fontsize=10, fontweight='bold')

ax1.set_xlabel('Eye radius (mm)', color='white', fontsize=11)
ax1.set_ylabel('Angular resolution (arcmin)', color='white', fontsize=11)
ax1.set_title('Compound-eye resolution vs size (Snyder 1979)', color='#f0abfc', fontsize=12)
ax1.set_xscale('log'); ax1.set_yscale('log')
ax1.legend(facecolor='#1a0a2e', edgecolor='#c084fc', labelcolor='white', fontsize=10)
ax1.grid(True, alpha=0.25, color='#a78bfa', which='both'); ax1.tick_params(colors='white')
for sp in ax1.spines.values(): sp.set_color('#c084fc')

# Panel 2: number of ommatidia
ax2 = axes[1]
ax2.set_facecolor('#1a0a2e')
ax2.semilogy(R_eye*1000, N_ommatidia, color='#f472b6', linewidth=2.5, label='Snyder optimum')

# Real species
N_species = [
    ('Fruit fly', 0.3, 700, '#fbbf24'),
    ('Bee', 1.4, 6000, '#f472b6'),
    ('Dragonfly', 2.5, 28000, '#22d3ee'),
    ('Mantis shrimp', 3.0, 16000, '#86efac'),
]
for name, r, n, c in N_species:
    ax2.scatter(r, n, c=c, s=140, edgecolors='white', linewidth=0.8, zorder=5)
    ax2.annotate(name, (r, n), xytext=(r*1.15, n*1.25), color=c, fontsize=10, fontweight='bold')

ax2.set_xlabel('Eye radius (mm)', color='white', fontsize=11)
ax2.set_ylabel('Number of ommatidia', color='white', fontsize=11)
ax2.set_title('Compound-eye pixel count vs size', color='#f0abfc', fontsize=12)
ax2.legend(facecolor='#1a0a2e', edgecolor='#c084fc', labelcolor='white', fontsize=10)
ax2.grid(True, alpha=0.25, color='#a78bfa', which='both'); ax2.tick_params(colors='white')
for sp in ax2.spines.values(): sp.set_color('#c084fc')

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#030712')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

4.2 Bat Echolocation: Biological Sonar

Microchiropteran bats (and odontocete whales, some birds) emit ultrasonic calls (20–200 kHz) and infer target range from time-of-flight, target size from echo amplitude, and target velocity from Doppler shift. Spallanzani (1794) first demonstrated that blindfolded bats fly normally while bats with plugged ears cannot — a century before Griffin (1944) identified ultrasound.

Derivation: Radar equation for bats

The received echo level at the bat's ear is:

\( RL = SL - 40\log_{10} r - 2\alpha r + TS \)

where \(SL\) is source level, \(\alpha\) atmospheric absorption coefficient (dB/m, strongly frequency-dependent), and \(TS\) target strength in dB. The high-frequency trade-off: higher \(f\) gives better spatial resolution\(\Delta r = c/(2B)\) but shorter range (absorption scales\(\sim f^2\)).

Doppler-shift compensation

Horseshoe bats (Rhinolophus) use constant-frequency (CF) calls with a narrow auditory “fovea” tuned to 80 kHz. Flying at velocity \(v\) toward a stationary target, the echo returns Doppler-shifted by:

\( f_{\text{echo}} = f_0 \frac{c + v}{c - v} \)

The bat lowers its outgoing call frequency to compensate, keeping the echo in the auditory fovea regardless of flight speed — the earliest-discovered biological adaptive beamforming (Schnitzler 1968).

Biomimetic sonar

• Chirp radar / LFM sonar — the frequency-modulated calls of Myotis are the biological origin of pulse compression.
• Bat-inspired ultrasonic obstacle avoidance for autonomous drones (Wang et al. 2020).
• Acoustic target classification via convolutional networks trained on bat-style chirps (Muller 2015).

Python

script.py77 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# Bat echolocation range equation
# Sonar range: received level RL = SL - 2*TL + TS
# SL = source level (dB re 20 uPa @ 1 m)
# TL = transmission loss = 20 log10(r) + alpha*r  (alpha = atm absorption dB/m)
# TS = target strength

SL = 120         # dB source level (typical bat shout)
NL = 40          # ambient noise level
DT = 15          # detection threshold SNR

# Atmospheric absorption (kHz, dB/m)
freqs = np.array([10, 20, 40, 60, 80, 100, 150])
alpha = np.array([0.1, 0.3, 0.8, 1.6, 2.5, 3.5, 5.5])  # dB/m (approx, Griffin 1971)

r = np.linspace(0.1, 50, 400)

fig, axes = plt.subplots(1, 2, figsize=(15, 6))
fig.patch.set_facecolor('#030712')

# Panel 1: Received level vs range
ax1 = axes[0]
ax1.set_facecolor('#1a0a2e')
TS = -20  # moth target (dB)
colors = plt.cm.magma(np.linspace(0.2, 0.85, len(freqs)))
for f, a, c in zip(freqs, alpha, colors):
    TL = 20*np.log10(r) + a*r
    RL = SL - 2*TL + TS
    ax1.plot(r, RL, color=c, linewidth=2, label=f'{f} kHz')
ax1.axhline(NL + DT, color='#e879f9', linestyle='--', linewidth=1.5, label='Detection threshold')

ax1.set_xlabel('Range r (m)', color='white', fontsize=11)
ax1.set_ylabel('Received level (dB)', color='white', fontsize=11)
ax1.set_title('Echolocation: received level vs range, moth target', color='#f0abfc', fontsize=12)
ax1.legend(facecolor='#1a0a2e', edgecolor='#c084fc', labelcolor='white', fontsize=9, ncol=2)
ax1.grid(True, alpha=0.25, color='#a78bfa'); ax1.tick_params(colors='white')
ax1.set_xlim(0, 30)
ax1.set_ylim(-10, 100)
for sp in ax1.spines.values(): sp.set_color('#c084fc')

# Panel 2: Maximum detection range vs frequency
ax2 = axes[1]
ax2.set_facecolor('#1a0a2e')

# Solve for range where RL = NL + DT
# RL = SL + TS - 40 log10(r) - 2*alpha*r = NL + DT
# transcendental; numerical solve
def find_range(SL, NL, DT, alpha, TS):
    rr = np.linspace(0.1, 200, 10000)
    RL = SL + TS - 40*np.log10(rr) - 2*alpha*rr
    valid = np.where(RL > NL + DT)[0]
    return rr[valid[-1]] if len(valid) else 0

ranges_moth = [find_range(SL, NL, DT, a, -20) for a in alpha]
ranges_wall = [find_range(SL, NL, DT, a, 0) for a in alpha]

ax2.plot(freqs, ranges_moth, 'o-', color='#e879f9', linewidth=2.5, markersize=9, label='Moth (TS = -20 dB)')
ax2.plot(freqs, ranges_wall, 's-', color='#22d3ee', linewidth=2.5, markersize=9, label='Wall (TS = 0 dB)')

ax2.set_xlabel('Frequency (kHz)', color='white', fontsize=11)
ax2.set_ylabel('Max detection range (m)', color='white', fontsize=11)
ax2.set_title('Frequency trade-off: range vs resolution', color='#f0abfc', fontsize=12)
ax2.legend(facecolor='#1a0a2e', edgecolor='#c084fc', labelcolor='white', fontsize=10)
ax2.grid(True, alpha=0.25, color='#a78bfa'); ax2.tick_params(colors='white')
for sp in ax2.spines.values(): sp.set_color('#c084fc')

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#030712')

print('Typical bat detection range on a moth at 40 kHz:', ranges_moth[2], 'm')
print('Typical bat detection range on a wall at 40 kHz:', ranges_wall[2], 'm')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

4.3 Pit Vipers: Infrared Imaging

Pit vipers (Crotalinae), pythons, and boas detect infrared radiation from warm-blooded prey via specialised loreal pit organs. The pit contains a thermally suspended membrane (15 μm thick) with dense innervation by TRPA1-expressing neurons that respond to temperature changes as small as 0.001 K (Gracheva et al. 2010, Nature).

Derivation: Thermal sensitivity

A target at temperature \(T_t\) and distance \(r\) radiates Stefan-Boltzmann power per unit area:

\( P_{\text{rad}} = \sigma \varepsilon \big(T_t^4 - T_{\text{env}}^4\big) \)

Radiant flux arriving at the pit membrane (area \(A_{\text{pit}}\)):

\( \Phi = \frac{P_{\text{rad}} A_{\text{target}} A_{\text{pit}}}{4\pi r^2} \)

For a rat (0.1 m², 310 K) vs background (293 K) at 1 m range, the snake receives\(\sim 10^{-7}\) W — an exquisitely small signal. Photon-counting detector analogues (quantum-dot bolometers) replicate this sensitivity in compact form factors for search-and-rescue and defence applications.

Applications

• Uncooled microbolometer arrays (FLIR, Raytheon) for thermal night vision.
• Pyroelectric passive infrared (PIR) sensors for motion detection — biomimetic simplified.
• “Pit-organ-inspired” soft thermoresponsive polymers (MXene-based, Yang et al. 2021).

4.4 Fish Lateral Line: Flow Imaging Underwater

Fish possess a line of mechanosensitive neuromastsalong the head and flanks that detect local water velocity. The lateral line system allows schooling coordination, prey capture in darkness, obstacle detection, and detection of dipole sources (struggling prey produce a characteristic near-field flow).

Derivation: Dipole flow detection

A vibrating prey of amplitude \(A\) and frequency \(\omega\) creates a near-field dipole velocity:

\( \mathbf{u}(\mathbf{r}) = \frac{A \omega a^3}{r^3}(2\cos\theta\,\hat{\mathbf{r}} + \sin\theta\,\hat{\boldsymbol{\theta}}) \)

where \(a\) is the prey radius. Each neuromast samples\(\mathbf{u}\) at one location; the array collectively resolves the dipole position and orientation via an inverse problem. Yang et al. (2006) built a MEMS neuromast array for underwater vehicle sensing, inspired directly by blind cavefish.

Applications

• Artificial lateral line arrays for AUV station-keeping and wake detection (Klein & Bleckmann 2011).
• Soft piezoelectric hair sensors on biomimetic fish robots (Asadnia et al. 2013).
• Vortex-tracking algorithms (leader-follower schooling) translated into marine robot swarms.

4.5 Electroreception: Passive and Active

Sharks, rays, and chondrichthyan fishes possess ampullae of Lorenzini — jelly-filled canals in the snout that detect electric fields as weak as 5 nV/cm, the most sensitive field-detection known in biology. Weakly electric fish (Gymnotiformes, Mormyridae) go further: they generate their own electric field via an electric organ (modified muscle) and sense distortions caused by prey, mates, and objects — an active electrolocation system analogous to radar.

Derivation: Electric dipole from prey

A flatfish buried in sand still has beating heart and respiration, producing ionic currents modelled as an electric dipole with moment \(\mathbf{p}\). The electric field at distance \(r\):

\( \mathbf{E} = \frac{1}{4\pi\varepsilon_0}\left[\frac{3(\mathbf{p}\cdot\hat{\mathbf{r}})\hat{\mathbf{r}} - \mathbf{p}}{r^3}\right] \)

Falling as \(r^{-3}\), detection is short-range (< 0.5 m) but directional. The shark's array of ampullae (up to 1800 on the snout) solves a 3D dipole-localisation inverse problem in real time.

Active electrolocation

Gnathonemus petersii emits electric organ discharges (EODs) at 4–7 Hz. Conductive objects (animals, metal) brighten the self-produced field; insulating objects (rocks, plastic) darken it. The animal forms an “electric image” on its skin electroreceptor array analogous to a retinal image. This has directly inspired underwater metal-detection sensors and biomimetic “electric catheter” devices for navigating blood vessels.

Applications

• Shark-inspired marine metal/cable detectors (e.g. for submarine wire inspection).
• Short-range underwater communication at frequencies around bat / fish EOD (Johnson et al. 2015).
• Bio-inspired active-electro AUV navigation where vision and sonar fail (Festo's AquaJelly).

Python

script.py76 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# Electric sense: shark (passive electroreception via Ampullae of Lorenzini)
# and electric fish (active electroreception via electric organ discharge)

# Dipole field: E = (p/4pi eps0 r^3) * (2 cos theta r_hat + sin theta theta_hat)
# Magnitude: |E| = (p/4pi eps0 r^3) sqrt(1 + 3 cos^2 theta)

X, Y = np.meshgrid(np.linspace(-3, 3, 200), np.linspace(-3, 3, 200))
r2 = X**2 + Y**2
r = np.sqrt(r2)
theta = np.arctan2(Y, X)

# Dipole along x: charges +q at d/2, -q at -d/2
# Total field magnitude (dimensionless)
p = 1  # dipole moment
# avoid singularity at origin
r_safe = np.maximum(r, 0.1)
E_mag = p / r_safe**3 * np.sqrt(1 + 3*np.cos(theta)**2)
E_x = p / r_safe**3 * (3*np.cos(theta)**2 - 1)
E_y = p / r_safe**3 * 3*np.sin(theta)*np.cos(theta)

fig, axes = plt.subplots(1, 2, figsize=(15, 6))
fig.patch.set_facecolor('#030712')

# Panel 1: Field magnitude + streamlines
ax1 = axes[0]
ax1.set_facecolor('#1a0a2e')
im = ax1.imshow(np.log10(E_mag + 1e-6), origin='lower', extent=[-3,3,-3,3], cmap='magma', alpha=0.9)

# Streamlines
ax1.streamplot(X, Y, E_x, E_y, color='#e879f9', linewidth=0.8, density=1.3, arrowsize=1)

# Dipole location
ax1.scatter([-0.3], [0], s=180, c='#22d3ee', edgecolors='white', zorder=5, label='+q')
ax1.scatter([0.3], [0], s=180, c='#f472b6', edgecolors='white', zorder=5, label='-q')

# Shark head silhouette at distance
ax1.plot([1.8], [1.5], marker='2', markersize=45, c='#fbbf24', zorder=6)
ax1.annotate('Shark ampullae\n(electric sense)', xy=(1.8, 1.5), xytext=(2.0, 2.3),
             color='#fbbf24', fontsize=9, fontweight='bold')

plt.colorbar(im, ax=ax1, label='log10(|E|)')
ax1.set_xlabel('x (body lengths)', color='white')
ax1.set_ylabel('y (body lengths)', color='white')
ax1.set_title('Dipole field from prey (flatfish cardiac)', color='#f0abfc', fontsize=12)
ax1.legend(facecolor='#1a0a2e', edgecolor='#c084fc', labelcolor='white')
ax1.tick_params(colors='white')
for sp in ax1.spines.values(): sp.set_color('#c084fc')

# Panel 2: sensitivity vs distance
ax2 = axes[1]
ax2.set_facecolor('#1a0a2e')
r_range = np.linspace(0.01, 2, 400)
E_axis = p / r_range**3 * 2    # maximum (along axis)

# Measured thresholds (shark 5 nV/cm ~ 5e-7 V/m, electric fish ~ 0.1 uV/cm = 1e-5 V/m)
ax2.loglog(r_range*100, E_axis, color='#e879f9', linewidth=2.5, label='Dipole axial field (arb units)')
ax2.axhline(5e-7, color='#22d3ee', linestyle='--', linewidth=1.5, label='Shark threshold 5 nV/cm')
ax2.axhline(1e-5, color='#fbbf24', linestyle='--', linewidth=1.5, label='Electric fish threshold')

ax2.set_xlabel('Distance (cm)', color='white', fontsize=11)
ax2.set_ylabel('|E| field (V/m)', color='white', fontsize=11)
ax2.set_title('Electric sense thresholds', color='#f0abfc', fontsize=12)
ax2.legend(facecolor='#1a0a2e', edgecolor='#c084fc', labelcolor='white', fontsize=9)
ax2.grid(True, alpha=0.25, color='#a78bfa', which='both')
ax2.tick_params(colors='white')
for sp in ax2.spines.values(): sp.set_color('#c084fc')

plt.tight_layout()
plt.savefig('output.png', dpi=150, bbox_inches='tight', facecolor='#030712')

Click Run to execute the Python code

Code will be executed with Python 3 on the server

4.3a Box Jellyfish: Eye Evolution Without a Brain

Box jellyfish (Cubozoa) possess 24 eyes arranged on four sensory structures (rhopalia), including two camera-type eyes with lenses. They detect obstacles and navigate to specific mangrove-root habitats — all without a centralised brain, using a distributed neural net. This offers a biomimetic model for extreme edge-computing visual navigation in highly constrained platforms. Nilsson et al. (2005, Nature) showed the upper lens eye is focused to only 4 deg sharp — deliberately blurred — which eliminates fine detail and reduces the computational load.

Lessons for biomimetic robotics: task-specific optical blur as a form of physical computation offload, an approach echoed in modern “simple eyes” on MAVs for obstacle avoidance without deep processing.

4.4a Seal Whiskers: Hydrodynamic Trail Following

Harbour seal (Phoca vitulina) whiskers have an undulated cross-section (oval with periodic bumps, Hanke et al. 2010) that suppresses vortex-induced vibration while maintaining high sensitivity to external wake patterns. A seal can follow the hydrodynamic trail of a swimming fish for up to 30 s after the prey has passed. Each individual vortex in the fish wake produces a detectable whisker deflection, and the array of whiskers resolves trail direction.

Engineering application: whisker-cross-section undulation has been copied to suppress vortex-induced vibration of marine risers, reducing fatigue damage on offshore oil platforms by factors of 2–5 (Beem & Triantafyllou 2015, J. Fluid Mech.).

4.5a Spider Slit Sensilla and Trichobothria

Spiders possess two spectacularly sensitive mechanosensors: slit sensilla (exoskeletal strain gauges) and trichobothria (long air-current-sensitive hairs on the legs). Trichobothria respond to airflows as gentle as 0.03 mm/s — sensitive enough to detect the wingbeats of a fly at 60 cm distance.

\( \theta_{\text{hair}}(\omega) = \frac{F(\omega) / S}{K_{\text{spring}} - J\omega^2 + i R\omega} \)

Hair deflection follows a damped driven oscillator response. Critical limit: the cricket cercal hair achieves detection of near-thermal-noise signals (\(\sim 10^{-20}\) J per event) — nature's most sensitive mechanical sensor. Biomimetic MEMS hair-sensor arrays (Izadi et al. 2010) replicate this using micromachined SU-8 cantilevers and piezoresistive readout.

Full treatment in Spider Biophysics Module 5.

4.5b Moth Pheromone Detection: Single-Molecule Sensitivity

Male silk moths (Bombyx mori) detect female pheromone (bombykol) at concentrations of one molecule per 10¹⁷ air molecules — demonstrably single-molecule sensitivity (Kaissling & Priesner 1970). The feathery antennae capture airborne molecules via a network of 17,000 olfactory sensilla, each with bound pheromone-binding proteins that funnel ligands to olfactory receptor neurons.

\( P_{\text{detect}} = 1 - \exp(-\sigma n \ell) \)

where \(\sigma\) is the capture cross-section, \(n\) the molecular density, \(\ell\) the antenna path length. Adkins-inspired electronic noses (Lewis group 2000, NeOse) aim for similar selectivity at room temperature using arrays of cross-reactive polymer films and pattern-recognition software.

4.6 Magnetic Sense: Birds and Radical-Pair Mechanism

Migratory birds navigate using the geomagnetic field via two candidate mechanisms: (1) magnetite crystals in the beak (compass needle model), and (2) cryptochrome-based radical pairsin the retina whose spin dynamics are influenced by the geomagnetic field (Ritz et al. 2000,Biophys. J.).

\( \Delta\Phi_{\text{singlet}}(\theta) = \cos^2\theta \cdot P_s^{\text{ref}} \)

The radical-pair mechanism is the first experimentally supported example of macroscopic quantum biology. Applications: quantum-sensing magnetometers (nitrogen vacancy diamond, 100 pT/√Hz) draw conceptually on the avian singlet-triplet population transfer.

Cross-reference: Quantum Biologycourse for detailed treatment.

4.7 Summary and References

Biology's sensors span the entire electromagnetic spectrum plus mechanics, chemistry, and hydrodynamics. Three architectural themes recur:

• Arrays of simple elements (ommatidia, ampullae, neuromasts) beat monolithic detectors at wide-field and robust operation.
• Adaptive gain / Doppler tracking (horseshoe bats) solves problems that fixed-gain engineering sensors cannot.
• Neural preprocessing at the sensor (pit-organ TRPA1 membrane, retinal ganglion motion detectors) offloads computation from central nervous systems — the inspiration for sensor-fusion neuromorphic chips.

References

[1] Snyder, A.W. (1979). Physics of vision in compound eyes. In Handbook of Sensory Physiology, vol. VII/6A. Springer.
[2] Song, Y.M. et al. (2013). Digital cameras with designs inspired by the arthropod eye. Nature 497, 95–99.
[3] Floreano, D. et al. (2013). Miniature curved artificial compound eyes (CurvACE). PNAS 110, 9267–9272.
[4] Griffin, D.R. (1958). Listening in the Dark. Yale University Press.
[5] Schnitzler, H.-U., Kalko, E.K.V. (2001). Echolocation by insect-eating bats. BioScience 51, 557–569.
[6] Gracheva, E.O. et al. (2010). Molecular basis of infrared detection by snakes. Nature 464, 1006–1011.
[7] Coombs, S., Gorner, P., Munz, H. (eds.) (1989). The Mechanosensory Lateral Line. Springer.
[8] Kalmijn, A.J. (1982). Electric and magnetic field detection in elasmobranch fishes. Science 218, 916–918.
[9] Bullock, T.H., Hopkins, C.D., Popper, A.N., Fay, R.R. (eds.) (2005). Electroreception. Springer.
[10] Ritz, T., Adem, S., Schulten, K. (2000). A model for photoreceptor-based magnetoreception in birds. Biophys. J. 78, 707–718.
[11] Asadnia, M., Kottapalli, A.G.P. et al. (2013). Artificial lateral-line system for underwater vehicles. J. R. Soc. Interface 12, 20150220.
[12] Shimozawa, T. et al. (2003). Cricket wind receptors: thermal noise for the highest sensitivity known. Trans. Am. Math. Soc. (Mechanoreception section).

← Module 3: Locomotion & Flight Module 5: Materials & Self-Assembly →