CrustBot + Grime Ward · tools for care of public space
Public space gets neglected in the gap between full sanitation crews and doing nothing — corridor edges, plaza spillover, park hotspots, drain approaches. Places too small or too frequent to clean often enough at conventional cost.
We’re building two products for that work. CrustBot develops affordable, remotely operated hardware for cleanup. Grime Ward helps people discover places, keep photo journals and contribute cleanup evidence. Follow the robot build or explore the app’s beta.
A simpler vehicle for the work ahead.
Current V1 simulation · uncalibrated dynamics
Remote drive, onboard camera perception, a powered jaw and a removable bin. The assembly packet is ready; physical commissioning comes next.
Places worth noticing. Visits worth keeping.
Current web beta · public-place discovery
Discover public places, keep dated photo visits and return to the same viewpoint. Share litter observations and completed cleanups when you choose.
Grime Ward grows from Trash Trail's litter-and-water map into a place journal. Discover nearby public places, keep your own dated photographs and return to the same viewpoint. Share litter observations and completed cleanups when you choose.
Explore mapped public places and nearby events. Check local access and source dates; coverage varies.
Keep private photos on this device, compare repeat viewpoints and export a backup. Journals are not account-synced.
Explicitly share litter photos, help identify observations and record completed cleanup work with dates and organization credit.
Remote driving, Pi 5 perception, Pico 2 control, a powered load-bearing jaw and a removable bin. The current design uses catalog drive parts, timber, plywood and selected printed mounts.
Discover public places, document repeat visits and share litter observations and completed cleanups. Private device journals and the public observation commons have separate consent.
Use measured access, effort and before-and-after evidence to develop repeatable cleanup work. Service routes, automatic pickup and robot deployment still need their own trials.
The first vehicle design includes remote control, Pi/camera perception, a powered jaw and a removable bin. Its plywood-and-timber frame has a checked assembly packet and a $604.29 planning material estimate. Active trash pickup and transfer come later. Physical commissioning, load tests and hot-weather endurance are still ahead.
October 15 build status and limits →
The failure mode isn't "never cleaned." It's "not cleaned often enough at the price point it would cost to do conventionally."
Neglected physical space communicates neglect back to the people living around it.
Customers need evidence, not vibes. Every pickup is paired before / after. Every benchmark is reproducible.
Waste robotics only becomes real if the dirty parts are survivable and replaceable.
Reviewed October 20, 2026. Robot engineering checkpoint: October 15; app release: October 9.
| Grime Ward · web beta | Discover public places, keep private photo visits on this device, revisit viewpoints, and explicitly share litter observations and cleanup records. Geographic coverage and evidence availability vary. |
| Grime Ward · iOS | Version 1.0, build 9, delivered to the internal TestFlight group on October 9. Public App Store availability is not yet established. |
| Initial vehicle | Remote driving, Pi 5 / IMX219 perception, powered load-bearing jaw and removable bin. Active litter pickup and transfer are deferred. |
| Frame assembly | CAD, dimensioned workshop drawings, four-page bench packet and eight P1S print projects are prepared. Unpowered frame assembly can begin from the instructions. |
| Planning material cost | $604.29, including $11 in unresolved fitting/price allowances. Owned material is counted; freight, tax, tools and labor are separate. |
| Perception | Desktop checks connect the recovered litter model to the read-only operator preview. Actual camera/Pi commissioning and useful field detection remain unverified; earlier benchmarks are separate evidence. |
| Simulation | V1 simulation lab · Current CAD, recorded Newton runs, interactive MuJoCo physics and two mapped real sites. Dynamics and terrain remain uncalibrated; this does not qualify field operation or six-hour endurance. Assembly review. |
| Still to verify | Delivered-part fit, wiring and stop behavior, loaded movement and jaw holding, bin/battery swaps, hot-weather operation and useful runtime. |
| Field pickups | 0 no chassis on a route yet |
The current CAD, workshop drawings and assembly sequence support an unpowered frame build. Delivered-part fitting and powered tests follow. Meet the current V1 →
The V1 simulation uses the current geometry and mapped sites. The next evidence is physical: traction, loaded stopping, jaw holding, temperature and useful runtime.
Web beta and internal TestFlight support discovery, private visits and shared contributions. Real-account sign-in, physical-device camera use and field usability remain part of testing.
DregsBane Systems LLC is a Virginia company, formed April 2026, building infrastructure against physical neglect: CrustBot, a remotely operated vehicle being prepared for assembly, and Grime Ward, a place journal and cleanup companion in beta.
It is deliberately small — everything on this site was designed, built, benchmarked, and deployed in-house. The build log is the honest record of who we are: what shipped, what didn't, and what the numbers actually say.
Contact: hello@dregsbane.com