Automated Robotic Pack Disassembly: What Battery Recycling Can Learn from White Goods

Most EV battery packs are disassembled by hand today: a technician discharges the pack, then works through fasteners, wiring harnesses, adhesive joints and potting compound one pack at a time. It’s slow, and it puts a person directly next to a charged pack that may be several hundred volts. Recycling Processes and Recycling Process Engineering cover what happens to a pack once it’s broken down. This post looks at the step before that. Getting it apart in the first place and specifically at vision-guided robotic disassembly, an area where battery recycling is following a path other industries have already started down.

The Precedent: Robotic Disassembly of Household Appliances

Automated end-of-life disassembly isn’t new, it’s just newer to batteries than to other consumer durables. Panasonic’s R&D centre in Osaka runs six-axis robots that disassemble air conditioners, microwave ovens and washing machines: magnetic grippers lift outer casings, machine vision identifies and locates fasteners (including screws obscured by dirt or rust, which a fixed program can’t handle but a vision model trained on many real units can), and a screwdriver end-effector removes them before other tooling cuts wiring and extracts components (ASSEMBLY Magazine, June 2026). The same programme feeds disassembly data back into product design. Panasonic redesigned the heat pump position in its washing machines specifically to make it easier for a robot (and a human repair technician) to access, cutting repair time by half. That closed loop between “how easy is this to take apart” and “how do we design the next one” is the more interesting long-term lesson for packs: disassembly-friendliness is a design decision made years before end-of-life, not something bolted on afterward.

The common technical thread across white-goods and electronics disassembly robots is the same: a general-purpose vision system trained to identify part type, condition and location on a product that varies unit to unit (different models, different wear, different residual dirt or corrosion), directing a manipulator that adapts its tool and force rather than following one fixed program per SKU.

What’s Already Been Done on Battery Packs

Oak Ridge National Laboratory built one of the earliest demonstrations specifically for EV packs, under the DOE’s Critical Materials Institute. Their system removes bolts and housing fasteners regardless of remaining charge, so a robot (not a person) is the last thing to touch a live pack before it’s in pieces. ORNL estimated the automated line could get through roughly 100 packs in the time a manual process handles about 12 (ORNL, 2021). More recent work has pushed further into perception: ORNL’s RAPID platform pairs a gantry-mounted industrial arm with RGB-D depth cameras and an automated nut-runner to identify and remove fasteners on a full-scale pack without a pre-programmed disassembly sequence, and a 2026 study demonstrated a vision-guided dual-arm system specifically for end-of-life 18650 cylindrical packs, using perception to locate and separate individual cells rather than just unbolting a housing.

Where the Analogy Breaks Down

Two things make battery packs harder than a washing machine, and they’re exactly the two the original gap in this topic called out — vision-guided de-potting and de-gluing:

  • Structural adhesive and potting compound. A washing machine’s casing comes off with screws. A modern pack’s cells or modules are frequently bonded in with structural adhesive for stiffness, vibration damping and thermal contact — removing them non-destructively means cutting or softening a bond line without a hard fastener to unscrew, which current commercial robotic disassembly (including the white-goods examples above) mostly doesn’t yet address at pack scale.
  • Cell-to-pack architecture removes the module step entirely. A pack built around discrete modules gives a robot an obvious intermediate disassembly target. Cell-to-pack (CTP) designs skip that level — cells sit directly in the pack structure — which removes a natural stopping point and means a disassembly system has to plan for direct cell-level extraction from the outset, a harder perception and manipulation problem than pulling out a bolted-in module.

On top of that, a failed appliance is electrically dead; a spent pack usually isn’t, and can still deliver a lethal shock or, if a cell is damaged, go into thermal runaway during handling — which is why every battery-specific system described above is built around minimising or eliminating direct human contact with a charged pack, a constraint the appliance precedent simply doesn’t have to design around.

References

  1. Weber, A. Appliance Manufacturer Uses Robots and AI to Disassemble Products, ASSEMBLY Magazine, June 2026
  2. Automated Disassembly Line Aims to Make Battery Recycling Safer, Faster, Oak Ridge National Laboratory / DOE Critical Materials Institute
  3. Robotised Disassembly of Electric Vehicle Batteries: A Systematic Literature Review, Robotics and Computer-Integrated Manufacturing, 2024
  4. Vision-Guided Dual-Arm Humanoid Robotic Disassembly of End-of-Life 18650 Lithium-ion Battery Packs, 2026

See also: Recycling Processes, Recycling Process Engineering, Recycling Companies

Leave a Comment