Every component a robot needs, needs many decisions.
Thirteen subsystems stand between a concept and a humanoid that actually stands up — and each hides a fork in the road. Pick the wrong motor, reducer or hand and you pay in torque, in grams, in dollars, and in months of lead time. roboBOM lays out every option against the constraints that decide it, so the trade-off is obvious before the purchase order goes out.
the anatomy of a humanoid
Where every system lives on the body
Hover a system to light it up on the robot — or hover the robot to surface the systems inside it. Click any row to jump to its options.
●●● cost & supply pressure★ what most humanoids ship◄ scroll any table right for suppliers · lead time ►
01
Joint actuator
how each joint is driven · logistics: heavy; rare-earth export rules
Type / option
Used when
Torque density
Backdrivability
Efficiency
Control ease
Cost
Suppliers
Lead
Quasi-direct-drive (QDD)★
legs & arms — dynamic, impact-tolerant joints
high
excellent
high
easy
●●●
Unitree, MIT-derived, in-house
8–14 wks
Geared electric (harmonic)
precision arms, wrists, waist
very high
poor
medium
precise
●●●
Harmonic Drive-based
12–20 wks
Series-elastic (SEA)
safe human-contact joints
medium
good
medium
force-friendly
●●●
in-house
10–16 wks
Hydraulic
legacy high-power (early Atlas)
very high
poor
low
hard
●●●
BD-style, custom
custom
›
Joint actuatorschematic
02
BLDC servo motor
the electromagnetic core · logistics: magnet content → export controls
Type / option
Used when
Low-speed smoothness
Torque density
Maintenance
Humanoid fit
Cost
Suppliers
Lead
BLDC, sensor-based (Hall)★
every powered joint
excellent
high
none
✓
●●●
T-Motor, Robotis
6–12 wks
BLDC, sensorless
fans & pumps — not joints
poor at low speed
high
none
✗
●●●
generic
4–8 wks
Brushed DC
toys, low-cost auxiliaries
good
medium
brushes wear
✗
●●●
many
2–6 wks
Stepper
3D printers, camera rigs
detented
medium
none
✗
●●●
OMC, generic
2–6 wks
Coreless
fingers & dexterous hands
very smooth
low
none
hands only
●●●
Maxon, Faulhaber
8–14 wks
›
BLDC servo motorschematic
03
Reducer
high-ratio gearing at the joint · logistics: fragile, precision handling
Type / option
Used when
Backlash
Torsional stiffness
Ratio range
Shock tolerance
Cost
Suppliers
Lead
Strain-wave (harmonic)★
arms, wrists — precision
near-zero
high
high
medium
●●●
Harmonic Drive, Leaderdrive
12–20 wks
Cycloidal
hips, knees — high load & shock
low
very high
high
very high
●●●
Nabtesco, Sumitomo
12–20 wks
Planetary
grippers, low-cost joints
moderate
high
low–med
high
●●●
Neugart, Apex
6–12 wks
›
Reducerschematic
04
Bearing
carries the joint load · logistics: standard freight
Type / option
Used when
Radial load
Axial load
Moment load
Compactness
Cost
Suppliers
Lead
Deep-groove ball
light auxiliary shafts
✓
light
✗
good
●●●
SKF, NSK
4–8 wks
Cylindrical roller
heavy radial shafts
high
✗
✗
medium
●●●
SKF, NSK
4–8 wks
Needle
compact wrist / finger
high
✗
✗
very compact
●●●
INA, IKO
4–8 wks
Thrust
vertical load points
✗
✓
✗
flat
●●●
SKF
4–8 wks
Crossed-roller★
every major joint
✓
✓
✓
very compact
●●●
THK, IKO
6–12 wks
›
Bearingschematic
05
Linear screw
rotary → linear motion · logistics: standard freight
Type / option
Used when
Efficiency
Load capacity
Precision
Self-locking
Cost
Suppliers
Lead
Ball screw★
linear leg / ankle actuators
~90%
high
high
✗
●●●
NSK, PMI
8–14 wks
Planetary roller
high-force compact linear
high
very high
very high
✗
●●●
Rexroth, GSA
10–16 wks
ACME / lead
manual adjust, low-cost lifts
low
medium
medium
✓
●●●
many
2–6 wks
›
Linear screwschematic
06
Encoder
closes the servo loop · logistics: light / air
Type / option
Used when
Resolution
Robustness
Absolute
Cost class
Cost
Suppliers
Lead
Optical
lab / precision arms
very high
dust-sensitive
option
high
●●●
Renishaw
6–12 wks
Magnetic★
most robot joints
med–high
robust
✓
medium
●●●
AMS, iC-Haus
4–10 wks
Capacitive
compact joints
high
medium
✓
medium
●●●
CUI
4–10 wks
Inductive
dirty / industrial joints
medium
very robust
✓
medium
●●●
Renishaw
4–10 wks
›
Encoderschematic
07
Compute (SoC / GPU)
the robot’s brain · logistics: export controls
Type / option
Used when
AI throughput
Power
Toolchain
Availability
Cost
Suppliers
Lead
NVIDIA Jetson / Thor★
most humanoids’ brain
high
med–high
mature (CUDA)
constrained
●●●
NVIDIA
8–20 wks
Tesla FSD SoC
Tesla Optimus only
high
efficient
closed
in-house
●●●
Tesla
n/a
Qualcomm / edge SoC
lighter cobots
medium
low
growing
good
●●●
Qualcomm
8–14 wks
Custom ASIC
high-volume production
tuned
lowest
build-it
NRE-heavy
●●●
fabless
6–12 mo
›
Compute (SoC / GPU)schematic
08
Battery pack
energy + power delivery · logistics: hazmat (UN38.3), duty
Type / option
Used when
Energy density
C-rate
Safety
Cost
Cost
Suppliers
Lead
Li-ion NMC★
most mobile humanoids
high
high
medium
medium
●●●
LG, Samsung SDI
6–12 wks
LFP
safety-first, stationary
medium
medium
high
low
●●●
CATL
6–12 wks
Li-Po
lightweight prototypes
high
high
low
medium
●●●
Molicel, many
4–8 wks
Solid-state
next-gen (future)
very high
medium
high
very high
●●●
emerging
R&D
›
Battery packschematic
09
Structure / frame
the load-bearing skeleton · material choice · logistics: bulky; Mg 85–90% China
Type / option
Used when
Strength : weight
Machinability
Fatigue
Supply risk
Cost
Suppliers
Lead
Aluminum★
main frame & housings
high
excellent
cracks
low
●●●
CNC shops
4–8 wks
Steel
shafts, gears, fasteners
medium
good
excellent
low
●●●
CNC shops
4–8 wks
Titanium
critical joint carriers
very high
poor (galling)
good
medium
●●●
Ti mills
6–12 wks
Magnesium
mass-critical limbs
very high
thixomolding
good
high
●●●
Mg (China)
8–14 wks
Composite
long limb segments
highest
forming-limited
impact-sensitive
medium
●●●
layup shops
8–16 wks
Polymer
covers & cable routing
low
excellent
good (damping)
low
●●●
injection molders
4–10 wks
›
Structure / frameschematic
10
PCBs & harness
power + signal distribution · logistics: light / air
Type / option
Used when
Movement zones
Current
Reliability
Cost
Cost
Suppliers
Lead
Rigid PCB
compute & power boards
static only
high
high
low
●●●
JLC, TTM
2–6 wks
Flex PCB
wrists, fingers, neck
flexes
medium
medium
medium
●●●
JLC, TTM
3–6 wks
Rigid-flex★
articulated joints
at joints
medium
high
high
●●●
TTM
4–8 wks
Wire harness
torso ↔ limb runs
anywhere
high
connector-dep.
low
●●●
Molex, TE
2–6 wks
›
PCBs & harnessschematic
11
Proprioceptive sensors
balance + force feedback · logistics: light / air
Type / option
Used when
Measures
Accuracy
Bandwidth
Cost
Cost
Suppliers
Lead
IMU (6/9-axis)★
torso balance
orientation, accel
med (drift)
high
low
●●●
Bosch, TDK
4–8 wks
6-axis F/T sensor
wrists & ankles
contact force/torque
high
high
high
●●●
ATI, Aidin
6–12 wks
Joint torque (strain)
each actuator
joint torque
medium
medium
medium
●●●
in-house
6–12 wks
Motor-current estimate
every motor (free)
rough torque
low
high
~free
●●●
firmware
n/a
›
Proprioceptive sensorsschematic
12
Tactile skin
touch sensing · logistics: light / air
Type / option
Used when
Spatial res
Durability
Maturity
Cost
Cost
Suppliers
Lead
Optical (GelSight)
fingertips (research)
very high
medium
research
high
●●●
GelSight
R&D
Piezoresistive
palms & grippers
medium
medium
some
medium
●●●
emerging
varies
Capacitive
hand surfaces
med–high
medium
some
medium
●●●
emerging
varies
Magnetic (Hall array)
robust fingertips
medium
robust
emerging
medium
●●●
Contactile
varies
›
Tactile skinschematic
13
End effector (hand)
manipulation · logistics: standard freight
Type / option
Used when
DoF
Dexterity
Payload
Cost
Cost
Suppliers
Lead
Parallel gripper
industrial pick-and-place
1
low
high
low
●●●
many
4–8 wks
Underactuated hand
general grasping
few
medium
medium
medium
●●●
in-house
8–12 wks
Tendon-driven dexterous★
human-like hands
many
high
low–med
very high
●●●
Shadow, Wonik
8–16 wks
Fully-actuated hand
research dexterity
many
high
medium
very high
●●●
in-house
12–20 wks
›
End effector (hand)schematic
Ranking reflects share of BOM cost, engineering difficulty and how often the part is the industry bottleneck. Options, suppliers and lead-times are indicative examples for comparison, not live quotes. Source: “Humanity’s Last Machine” — a component-level map of humanoid robotics hardware.
stop guessing. start speccing.
From a napkin to a BOM that actually bolts together.
Hand roboBOM your requirement or your enclosure. It matches every constraint, picks the part, and proves the fit in 3D.