Questions for Waste Management — Desktop Driving Safety Simulator Pilot

Answers are needed before Friday, because that is the day scope and pricing are fixed — after it, anything still open has to be guessed at, carved out, or re-quoted later.

The nineteen are numbered in order of how much each one holds up. The first three hold up the most. Within each section below, the numbering runs in that same order, so a low number is a question worth chasing first.

This is a list of questions, not a quote. Nothing here carries a price.

Answer these three first

Questions 1–3

These three sit underneath everything else. Until they are settled, most of the remaining answers can move.

Does “15–20” mean licensed drivers, or fifteen to twenty people training at the same time?

The pricing workbook gives four different sizes for the same pilot. The scope header says one to two desktop workstations across two training centres. The seat line says fifteen minimum, twenty preferred. The budget summary says eighteen seats. Every hardware line is quantity two. And one of the questions asks us to confirm the platform has been tested at fifteen to twenty simultaneous sessions.

Which is it — fifteen to twenty drivers licensed to use the system over the course of the pilot, or fifteen to twenty sessions running at the same moment?

Why the answer mattersTwo workstations cannot produce fifteen simultaneous sessions. If WM means twenty machines running at once, the licence count, the hardware, the reporting layer and the installation plan all change — that is a different project from the one described in the scope header.

Is the following gap four seconds or five? Is the curve rule ten mph below the posted limit, or fifteen?

The pricing workbook specifies a five-second following gap, and fifteen mph below the posted speed limit on curve entry. On the 8 July call, Jake Bell told us WM teaches a four-second minimum following distance, and that drivers are required to drop ten mph below the posted limit before they enter the curve.

Which pair does the simulator score against?

Why the answer mattersThis is the one item ForgeFX could plausibly guess and be quietly wrong about. A simulator scoring against a standard WM does not teach would hand your trainers a complete set of numbers that disagree with them.

Is the pilot backing only?

WM’s requirements document is backing-only across all six evaluation categories. WM’s pricing workbook also prices following distance and rollover, and asks for a force-feedback wheel for rollover training plus a commercial route environment to run it in.

Is the pilot backing only, or backing plus following distance plus rollover?

Why the answer mattersIt sets how many scenarios get built, which environments get built, whether the truck needs a physics model capable of actually rolling over, and whether following distance and rollover sit inside the pilot or are shown separately.

How performance is measured

Questions 4, 5, 6, 14, 15

Scoring rules have to come from WM. ForgeFX will not invent a WM safety standard, and a threshold we guess at is a threshold your trainers will disagree with.

Will WM accept an in-simulator mirror check instead of eye-tracking hardware?

On a desktop simulator, the reliable way to prove a driver checked a mirror is to have them press a button on the steering wheel to pull the mirror into view — a discrete, time-stamped, logged event. The alternative is gaze-tracking hardware, which on a desktop setup is fragile and adds cost.

WM’s own document hedges here — “eye-tracking capability (if hardware supports it)” — and Jake Bell named the problem directly: if you cannot detect eye movement, how do you measure mirror scanning? Will WM accept the in-simulator confirmation as the method of record?

Why the answer mattersA yes closes the only real technical risk in the programme. A no means WM funds the gaze hardware, and the scanning measures are built around it instead.

What is the 3-6-9 scanning process?

The term appears in WM’s requirements document and nowhere else — not in the pricing workbook, and in neither of the two calls. ForgeFX needs the sequence, what the intervals refer to, and what counts as a pass.

Why the answer mattersScanning is a scored behaviour in two of the six evaluation categories. ForgeFX will not guess at a WM safety standard, so this one blocks the scanning configuration outright.

What clearance distance counts as a near miss?

WM has given ForgeFX two precise standards — three mph maximum while backing, and a three-second pause — but no number for this, and “near misses” is a listed measure. Four parts travel together:

What clearance counts as a near miss?
Is that the same figure for every object type, or set per object?
Is contact an automatic fail, or a scored deduction?
Is clearance measured from the truck body, from the mirrors, or from the full swept path?

Why the answer mattersOn a refuse truck the mirrors and the raised container are the parts that hit things first, so where the measurement starts changes the result on the same drive.

Is a momentary excursion above three mph a violation, or must the driver hold above it?

WM lists both “maximum speed during manoeuvre” and “time spent above speed threshold” as measures, without saying which one drives pass or fail. A second part rides along: does backing-speed monitoring run only inside the designated backing exercise, or continuously across the whole session?

Why the answer mattersThe two readings produce different scoring behaviour, and can pass and fail the same driver on the same run.

Live in-session feedback, or post-session scoring only?

Greg Meyers put this choice to WM on 8 July — show the information live, as a countdown or on-screen readout, or track it quietly and review it afterwards — and it is still open.

Why the answer mattersIt changes the interface, not just the report. A guided mode teaches the standard while the driver drives; an unguided mode tests whether the driver has internalised it. They are different products.

How the backing exercise runs

Questions 7, 11, 12, 13

These describe the exercise itself — the course, the required steps, and what happens when a driver misses one.

Please send the backing course diagram, and the pass tolerance

WM asked ForgeFX to replicate the standard backing exercises performed on WM training courses. A generic lane is not a replica. ForgeFX needs:

Lane length, lane width, marker type and marker spacing.
The pass tolerance for “position within defined boundaries” — in inches or feet.
Whether contact with a marker is a deduction or a fail.
Whether the exercise ends at a fixed target distance, or when the driver judges the stop.

Why the answer mattersThis is the difference between building WM’s course and building a course. Without the diagram, drivers train on a layout they will never see again.

Is a horn tap required before backing — how many, and at what point in the sequence?

WM’s document writes “(if required)”, which leaves ForgeFX scoring a rule WM has not stated. The four-way flashers have the same problem: the horn and the flashers both appear for the first time in the requirements document and were raised in neither call.

Why the answer mattersIf these are graded behaviours, ForgeFX needs the rule to score them. If they are not, ForgeFX should not be building the check at all.

Does the three-second count start at gear selection, or at the completed shift into reverse?

In a loaded truck the two moments can be a noticeable distance apart, and this is the behaviour Jake Bell asked for by name: drivers place the truck in reverse and pause three seconds before they move, and he wanted the simulation to measure exactly that.

Why the answer mattersStarting the clock at the wrong moment quietly shifts every pause result in the pilot in one direction.

When a driver violates a step, does the exercise stop and reset, or continue and log it?

Either is workable, but ForgeFX needs to know which one WM’s trainers use today.

Why the answer mattersThis changes how the scenario is built, not only how it is scored — a stop-and-reset exercise and a run-to-completion exercise are constructed differently.

The truck and the setting

Questions 8, 16

What ForgeFX models, and what reference material WM can put in front of the team.

Which refuse truck is the pilot vehicle, and what reference material can WM provide?

Four parts:

Which specific chassis and body?
What reference can WM give — CAD, a photo-and-measurement session at a facility, or physical access to a vehicle?
Is the pilot truck front-load, as the pricing workbook says, or a residential collection truck, as the backing content in the requirements document implies?
Automatic or manual — does the workstation need a clutch pedal?

Why the answer mattersGreg Meyers named vehicle access as a standing project risk on 8 July: without the real equipment, the model is a likeness rather than a match. The transmission answer also decides the control hardware.

Phoenix or Tampa for the residential environment?

ForgeFX needs one of the two named so the team builds the right signage, vegetation, housing stock and lot geometry.

Why the answer mattersIt does not change the price. It does decide whether the environment looks like the streets your drivers actually work.

Systems, data and connectivity

Questions 9, 10, 17, 18

Where results go, what the workstations can reach, and which WM systems they have to speak to.

Can WM grant Kahuna sandbox access, and supply Kahuna’s xAPI/LRS documentation?

ForgeFX has not integrated with Kahuna before. Two things are unknown today: whether Kahuna accepts custom xAPI statement verbs and extensions, and whether a sandbox tenant can be made available inside the project schedule. Who is the Kahuna administrator ForgeFX should be talking to?

Why the answer mattersWith sandbox access, ForgeFX builds and tests against the real system. Without it, ForgeFX delivers to a standard learning-record endpoint and WM’s Kahuna administrator owns the last mile. Those are two different pieces of work, and the choice is WM’s, not ForgeFX’s.

Are the training centres permanently connected, intermittently connected, or air-gapped?

Two further parts: what encryption standard does WM IT require for results held on the workstation between uploads, and does this component trigger a WM security review with its own timeline?

Why the answer mattersIf the centres are always online, WM may not need the offline queue at all. If they are air-gapped, remote monitoring cannot work as described and ForgeFX would substitute an on-demand diagnostic bundle instead.

How does WM define a cohort, and where do cohort attributes come from?

WM cites “new hire vs. refresher” without a full list. ForgeFX needs the cohort attributes and their source — Kahuna roster fields, or entered locally at the workstation. Two logistics items ride along: how many workstations at each centre, and the street addresses of the Arizona and Florida training centres.

Why the answer mattersReporting groups drivers by cohort, so the definition has to come from WM. The addresses and workstation counts set the installation plan.

Which telematics and dash-cam platform does WM run, and does it expose a documented API?

A single name and a yes or no on the API is enough to start.

Why the answer mattersThat one answer is the difference between a routine integration and an open-ended one. It cannot be scoped platform-independently.

Brand artwork and permissions

Question 19

Easy to answer, and easy to forget until the day it is needed.

Will WM supply brand artwork, and written permission to use WM marks?

ForgeFX needs decal and numbering specifications for the truck, and written permission to use WM marks in two places: inside the delivered application, and in ForgeFX demonstration material.

Why the answer mattersThe second half is a separate grant from the first, and it is the one that tends to surface late.

Nineteen questions only Waste Management can answer