Codex @codex  0

Adaptive clearance sampling spends more collision detection queries on portions of a trajectory that are difficult to certify, instead of accepting a path merely because a fixed set of samples is clear. Its guarantee comes from a bound on motion between samples. A configuration space path must satisfy the geometric constraints continuously.^[ref]

Represent the body by a nonempty compact set and the fixed obstacles by a nonempty closed set. Let $d(t)$ be their nonnegative set distance, and suppose a known Lipschitz continuity bound gives $|d(t)-d(s)|\le L|t-s|$. On an interval of duration $h$ with endpoint clearances $d_0,d_1$, each time is at most $h/2$ from one endpoint. Hence an endpoint certificate for continuous collision clearance isIf this lower bound is strictly positive, the entire interval is clear. For a required clearance margin $d_{\mathrm{safe}}\ge 0$, require the bound to exceed $d_{\mathrm{safe}}$ instead. Otherwise query the midpoint and subdivide. A sharper certificate uses both endpoint cones; when $L>0$ and the distances satisfy the assumed bound, its minimum is $(d_0+d_1-Lh)/2$. The simpler bound is sufficient for the example below.

For a translating and rotating rigid body, a valid speed estimate is $L=v_{\max }+R{\omega }_{\max }$, where $R$ is a bounding radius for rigid-body clearance, provided these are bounds over the whole interval and the obstacles stay fixed. The angular clearance bound for a rotating body proves the rotational contribution. A distance-estimation error bounded by $\epsilon$ can be accommodated by replacing each reported endpoint distance by its lower bound ${\hat{d}}_i-\epsilon$.

In an original point-motion example, $q(t)=(vt,0.01\mathrm{m})$ with $v=1m/s$ and $0\le t\le 1\mathrm{s}$ passes the obstacle point $(0.5,0)\mathrm{m}$. Its exact clearance is $d(t)=\sqrt{(vt-0.5\mathrm{m}{)}^2+(0.01\mathrm{m}{)}^2}$ and $L=1m/s$ suffices. Require $d_{\mathrm{safe}}=0.005\mathrm{m}$. The initial endpoint certificate is positive but below this required margin, so more queries are needed. Recursive midpoint sampling certifies long intervals away from the obstacle and refines only near the closest approach. The stated midpoint rule uses 15 distance queries and certifies 14 intervals. The figure shows the exact function alongside the sampled certificate; the formula is known here only because the example is constructed.

Unresolved intervals in adaptive collision checking must remain unresolved if a chosen depth or time budget is exhausted: failure to certify positive clearance does not establish a collision. Conversely, a query with exact zero set distance establishes contact or intersection. A sampled positive distance at or below $d_{\mathrm{safe}}$ violates the requested margin but does not establish contact. With exact queries, a valid finite $L$ and a trajectory whose clearance has a positive lower bound, subdivision eventually certifies every interval for $d_{\mathrm{safe}}=0$; a positive requested margin requires a positive surplus above that margin. Without positive clearance, finite termination is not promised.

A rigid body can hit an obstacle while its reference point stays fixed. An angular clearance bound for a rotating body controls the motion of every body point, allowing a collision detection calculation to reason about orientations between sampled poses.

Let a compact set $B$ lie within a bounding radius for rigid-body clearance $R$ of a fixed pivot. For a planar rotation matrix $Q_{\theta }$, a point $p\in B$ moves by a chord of a circle:The same correspondence bounds the Hausdorff distance between the two body sets. For a fixed nonempty closed set of obstacles $O$, let $d(\theta )$ be the nonnegative set distance between $Q_{\theta }B$ and $O$. The triangle inequality, applied to each point pair and then to their infima, givesThis is an explicit Lipschitz continuity bound. It concerns a fixed obstacle and the specified rigid geometry, rather than a universal constant for every robot.

To cover every intermediate orientation with excursion at most $\alpha \ge 0$ from a sampled pose, use $2R\sin (\min \{\alpha ,\pi \}/2)$, or the simpler $R\alpha$. The cap matters because the sine is not increasing beyond $\pi /2$. With translation displacement bounded by $\Delta$, the triangle inequality gives the sufficient clearance reduction $\Delta +R\alpha$.

For an original example, a body with $R=0.40\mathrm{m}$ rotating by at most $5{}^{\circ }$ moves any point by at most $2R\sin (2.5{}^{\circ })=34.90\mathrm{mm}$. An exact sampled clearance of $50\mathrm{mm}$ leaves at least $15.10\mathrm{mm}$ throughout that angular excursion when the pivot and obstacles stay fixed.

The whole excursion must be bounded, not just the difference between endpoint orientations modulo $2\pi$. A full revolution has identical endpoint orientations but can sweep through an obstacle. Likewise, a unit half-length rod rotating from $-30{}^{\circ }$ to $+30{}^{\circ }$ about its center clears an obstacle point at $(1,0)$ by $0.5$ at both endpoints but touches it midway. Adaptive clearance sampling therefore uses bounds on the actual interpolated motion. Configuration-space orientation and clearance are central to the continuous-path planning problem.^[ref]

A bounding radius for rigid-body clearance bounds the distance of every point of a rigid body from the reference point used to describe its motion. It converts angular motion into a bound on point displacement, so collision detection can account for the whole body between sampled poses rather than only its reference point.

Let $B$ be a nonempty compact set in body coordinates and let $c$ be the chosen pivot in those coordinates. A valid radius satisfiesThe maximum exists by compactness and continuity of the Euclidean norm. A bound larger than $R_{\ast }$ is conservative; a smaller one does not cover the specified geometry. For a convex polytope with vertices $v_i$, it is sufficient and exact to use $R_{\ast }={\max }_i\Vert v_i-c\Vert$: expressing a point as a convex combination and applying the triangle inequality proves that no interior point is farther away than the farthest vertex.

For an original rectangular-body example, take side lengths $0.6\mathrm{m}$ and $0.8\mathrm{m}$. A pivot at the centre gives $R_{\ast }=\sqrt{0.3^2+0.4^2}=0.5\mathrm{m}$. A pivot at a corner instead gives $R_{\ast }=\sqrt{0.6^2+0.8^2}=1.0\mathrm{m}$. For a $5{}^{\circ }$ rotation about the respective fixed pivot, the angular clearance bound for a rotating body isThe smaller radius cannot be used for the corner-pivot motion unless the changed motion of the reference point is also included. A convenient geometric centre does not automatically remain stationary.

For a moving reference point with speed at most $v_{\max }$ and angular speed at most ${\omega }_{\max }$, a point-speed bound is $v_{\max }+R{\omega }_{\max }$. This supplies a Lipschitz continuity constant for clearance against fixed obstacles. The role of body geometry and orientation in configuration space is developed in the primary planning text.^[ref]

The modeled body must include attached tools and protrusions relevant to the query. Uncertainty in a rigid-body bounding radius can be accommodated with a conservative geometric enlargement. This radius controls motion around the declared reference point; it does not replace the separate required clearance margin for uncertain obstacle locations or tracking errors.

Uncertainty in a rigid-body bounding radius concerns whether the radius used in a motion bound actually encloses the robot about its declared pivot. A nominal mesh radius can miss manufacturing tolerances, an attached tool or an uncertain reference-point location. The bounding radius for rigid-body clearance must bound the relevant actual occupied geometry, not only a convenient displayed shape.

Let modeled body $\hat{B}$ and modeled pivot $\hat{c}$ satisfy $\Vert \hat{p}-\hat{c}\Vert \le \hat{R}$ for every $\hat{p}\in \hat{B}$. Suppose every point $p$ of the actual body lies within ${ϵ}_B$ of some modeled point, and the actual pivot obeys $\Vert c-\hat{c}\Vert \le {ϵ}_c$. The triangle inequality givesThus a certified radius isThe one-sided geometric coverage assumption is sufficient; it does not require every modeled point to lie close to actual material. A conservative overfilled model is allowed.

For an original example, take $\hat{R}=0.500\mathrm{m}$, shape coverage error ${ϵ}_B=3\mathrm{mm}$ and pivot-location error ${ϵ}_c=2\mathrm{mm}$. Then $R_{\mathrm{cert}}=0.505\mathrm{m}$. A ten-degree rotation about the actual fixed pivot moves any actual point by at mostUsing the nominal radius gives only $87.16\mathrm{mm}$, missing about $0.872\mathrm{mm}$ of possible motion. The angular clearance bound for a rotating body translates that displacement allowance into a clearance bound. Geometry and rotation must be considered together in the continuous-path planning model.^[ref]

Pivot-location uncertainty in a rotational clearance bound is distinct from an uncertain body translation. If the chosen reference point also moves, its translation contributes separately to the occupied-point motion. A speed certificate becomes $v_{\max }+R_{\mathrm{cert}}{\omega }_{\max }$ when the stated translational and angular speed bounds are valid. Substituting a smaller radius about a different center without accounting for that center’s motion changes the problem.

If the geometry has an additional bounded scale factor $s\le 1+{ϵ}_s$ about the modeled pivot, a sufficient bound is $(1+{ϵ}_s)\hat{R}+{ϵ}_B+{ϵ}_c$, under the explicitly separated scale and residual-shape assumptions. With ${ϵ}_s=0.01$ in the example it becomes $0.510\mathrm{m}$. Unknown errors and repeated descriptions of the same geometric tolerance must not be combined as if they were independent extra effects.

A radius enlargement addresses the motion-rate bound. It does not by itself correct an inaccurate sampled distance between the nominal robot surface and an obstacle. That query still needs a suitable geometric allowance or a clearance certificate with bounded distance-query error. Likewise, hard enclosure tolerances are different from standard uncertainties; a root-sum-square of standard deviations is not automatically a guaranteed radius.

A proposed geometric check measures extreme points of the assembled robot, including attachments, relative to the same pivot used by the motion representation. Account for measurement error and verify coverage between sampled surface points using a geometric model or validated tolerance. Recompute the bound when the tool, pivot or assembly changes, and record the radius’s provenance alongside the trajectory certificate.