A G-force meter that does not care which way the phone is pointing.
Lateral, braking and acceleration G, computed from the GPS trace rather than a tilted accelerometer, shown live as a friction circle and stored with every corner of every drive.
Speed times the rate of heading change, divided by gravity. Reported for the trip as a peak and an average, and per corner as the G the corner was taken at.
Change in GPS speed over time. Decelerations past 3.5 m/s² are counted as brake events, and the hardest one is the trip's max deceleration.
The accelerometer's job. Vertical vibration, frequency-weighted the way ISO 2631 specifies, becomes the ride comfort grade and the road-quality index.
Why GPS beats the accelerometer for horizontal G
An accelerometer measures acceleration along the phone's own axes. In a car, the useful axes are the car's: forward, sideways, up. The two only line up if the phone is mounted level and square to the direction of travel, and even then gravity leaks into the horizontal readings the moment the road tilts. Every G-meter app that reads the accelerometer directly has a calibration step for exactly this reason, and it drifts as soon as the mount moves.
Torq sidesteps the problem. GPS reports speed and course once a second. The rate at which the course changes, multiplied by the speed, is the centripetal acceleration of the car, in the car's own frame, regardless of where the phone is. The change in speed from one fix to the next, divided by the time between them, is the longitudinal acceleration. Both are divided by 9.81 to give G. No calibration, no orientation, no gravity leak.
The cost is time resolution. A one-second sample cannot see a 200-millisecond spike, so a single sharp jolt registers lower than a racing data logger would show. For a corner that lasts three seconds or a stop that lasts two, it is more than enough, and the reading is a filtered one: Torq takes the 90th percentile of the per-sample G through a corner rather than the single highest sample, so one bad fix cannot invent a peak.
What the motion sensors do instead
The gyroscope confirms that a heading change is a real yaw of the car and not a GPS wobble, which matters at low speed where the course reading gets noisy. The accelerometer's vertical axis is filtered and weighted to produce the comfort and road grades described on the road-quality page. A phone drop filter throws away any sample above 3 G, since nothing a road car does on tires produces that.
The friction circle
The circle on the drive screen plots lateral load on the horizontal axis and braking or acceleration on the vertical. The dot is where the load sits now; the outer ring is roughly where road tires stop gripping. Trail-braking into a corner draws the dot from the top toward the side; a clean exit sweeps it from the side toward the bottom. It is the same diagram data engineers use to read a driver, drawn from what a phone can see.
Where the numbers end up
Peak G, average corner G, max acceleration and max deceleration are in the vehicle dynamics section of every trip. Each corner in the corner list carries its own lateral G alongside its entry, apex and exit speeds; the corner page shows how that becomes a rating. The all-time peak G is one of the personal records in the Ranks tab. The safety grade counts hard brakes and aggressive turns per 100 km, so a drive with many high-G moments scores lower there even if the corners were graded well.
Questions people ask
How does a phone measure G-force in a car?
Two ways, and Torq uses both. Lateral G in a corner is computed from the GPS trace: speed multiplied by the rate the heading is changing gives centripetal acceleration, and dividing by 9.81 gives G. Braking and acceleration G come from the change in GPS speed over time. The accelerometer and gyroscope are used for the vertical axis and for confirming that a heading change is a real turn.
Why not just read the accelerometer?
Because a phone in a mount is never perfectly level, and a phone in a pocket is worse. A raw accelerometer reading mixes the car's motion with gravity leaking into the wrong axis and with every bump. Deriving the horizontal G from GPS speed and heading removes the orientation problem entirely; the motion sensors then only have to answer questions they are good at, like how rough the road is.
What G-force is normal on the road?
Ordinary driving stays under about 0.3 G in every direction. A brisk corner on a good road is 0.4 to 0.6 G, and a hard stop in traffic is around 0.5 G. Road tires on a dry surface give up somewhere near 0.8 to 1.0 G, which is why Torq caps the peak it will believe from a road drive at 3.0 G and treats anything above as a sensor artifact.
What counts as a brake event or a hard acceleration?
A brake event is a deceleration of 3.5 m/s² or more, about 0.36 G, sustained rather than a single sample. A hard acceleration is 2.8 m/s² or more. Both thresholds come from telematics research on what drivers and passengers perceive as harsh, and both are counted per trip and shown in the vehicle dynamics section.
Does the phone need to be mounted?
For the lateral and longitudinal G, no: those come from GPS and work with the phone anywhere in the car. For the vertical axis, which feeds ride comfort and road quality, a mount matters, because a phone sliding around a seat measures the seat, not the road.
Is the peak G shown live?
The friction circle on the drive screen shows the current lateral and longitudinal load while recording. The peak G for the trip, the average corner G, and the G on each corner are computed when the drive ends and appear in the trip detail. G-force and cornering breakdowns are part of Torq Pro.
See the load, corner by corner.
Torq turns GPS speed and heading into lateral G, counts every hard stop, and keeps the peak for each drive.
iPhone, iOS 17 or later. No account. First drive free.