A pressure sensor must separate a user’s resting-force baseline from an intentional extra push

What it is

A thumb already pressing the bezel just to reach a key: most of the force the sensor sees is to hold the device, not a command. Baseline versus intentional force offset asks the system to estimate this hand’s current rest force, and to treat only the slice above that baseline as a command. How many levels can be held, and whether pressure gestures are discoverable, are neighboring problems. This layer asks where command force starts.

Why it happens

Rest force moves with grip, posture, gloves, and device weight: a thumb on the upper edge in a one-handed hold sits far above two hands on a desk. A single global threshold leaves light grips never reaching a second level, and death-grips crossing on contact. Baseline also drifts inside a press—people push harder to keep posture. So the estimate has to be local: a window right after contact, or a low-pass that tracks slow change and treats only a fast rising edge as command. Follow too tightly, and an intentional push is absorbed into baseline; too slowly, and a regrasping is a hard-press. Position and force share a finger, so aiming at the upper edge itself raises baseline—the reason edge targets false-trigger the force channel.

Studying it

Record force time series after contact under different grips and gloves, mark the “not yet pushing on purpose” window, and compare a fixed threshold with adaptive baselines.

Independent variables: grip (desk / one hand / walking), gloves, device weight and edge targets, baseline algorithm (fixed / contact window / slow follow). Dependent measures: rest-state false crosses, missed intentional pushes, regrasping counted as hard-press.

Do not only measure a finger pressing straight down in a fixture. Split “tap the upper edge” from “tap the center”; edge baseline rise is the main error. Adaptive algorithms should report the trials where they lost the baseline.

Where it stops holding

A stylus tip has stable contact area and the hand does not hold the device through the pen, so baseline is milder; pen pressure policy does not copy onto a trackpad. On a soft pad or a lap, whole-device sink is read as force and baseline jumps. Children and weak grips have a tiny command offset; an overestimated baseline leaves no remaining dynamic range. If a safety-critical “push a bit harder” rides this, a wrong baseline is a false trigger or an empty press; use a second stage of travel, not a force offset.

Applying it

• Threshold the force channel relative to a local post-contact baseline, not to a factory Newton number.
• Follow slower than an intentional rising edge and faster than a regrasping; if you cannot, disable the force channel during regrasping.
• Do not bind hard-press to edge targets that already take force to reach.
• Verify: desk, one hand, walking, target gloves; in each, “contact without pushing” and “push after contact.” Rest must not cross; pushes must not miss. Run a dedicated block of upper-edge taps. Baseline-loss segments must be retrievable from logs.

Related

• Same group: C10.12.2 Holding a pressure level continuously fatigues the hand more than holding a position · C10.12.3 How many force levels can be reliably told apart is limited by sensor precision and the stability of human force · C10.12.4 Without synced visual or haptic confirmation, people cannot tell which force level they have reached
• Adjacent: C10.04 Pressure Input · C5.09 Pressure Curves and Individual Fitting
• Search: force baseline · resting force · intentional offset