3.2 Route Planning and Coverage Strategy
The path you choose through a space is the most important technique decision after speed and posture. Good routes create loop closures, distribute coverage, and produce data that processes without complication. Poor routes accumulate drift and leave gaps that cannot be filled in post-processing.
Core Route Planning Principles
A scan route is not just a path. It is a sequence of decisions about where to create loop closures, where to provide redundant coverage, and where to transition between spaces. The principles below apply to every project regardless of scale or complexity.
Walk Main Corridors First, Branch Into Rooms Second
Establish the spine of the floor plan before branching into individual rooms. Walking the main corridors first gives SLAM a continuous trajectory backbone to anchor everything else against. Each subsequent room entry and exit becomes a loop closure relative to that established backbone, which corrects drift continuously throughout the session.
Branching into rooms before the corridor spine is established produces a session where the room-to-room relationships rely on individual transitions to maintain alignment. This is less robust and accumulates more uncorrected drift.
Each Room Entry and Exit Is a Loop Closure
When you enter a room from the corridor, scan the room completely, and return through the same doorway you entered, you create a loop closure at the doorway. The system has now seen the corridor twice (before and after the room), giving it the geometric reference needed to detect any drift that accumulated inside the room and apply a correction across that segment of trajectory.
Exiting through a different doorway than you entered loses this loop closure. It still scans the room, but the corrective comparison is weaker. Where multiple doorways exist, plan your route so each room is entered and exited through the same opening when possible.
Close the Final Loop at the End
The most important loop closure of any session is the final one: returning near the initialization point at the end of the scan. This loop closure corrects drift accumulated across the entire session, not just one segment. Always return to within 15 to 30 ft (5 to 10 m) of where you started, from a similar viewing angle.
Walking the full perimeter of a building and ending at the starting point from the opposite direction does not satisfy this requirement. The return viewing angle must be within 40 degrees of the original initialization angle.
Movement Patterns Within Spaces
Inside rooms and open areas, the path you walk determines coverage quality. Random movement leaves gaps; systematic patterns ensure even coverage with multiple angles on every feature.
Serpentine (S-Pattern)
Walk back and forth across the room in S-shaped sweeps, with each sweep offset from the previous one by approximately 6 to 10 ft (2 to 3 m). This produces dense, even coverage with the LiDAR observing each surface from multiple angles.
Lawn-Mowing
Walk parallel passes the length of the room, then return on the next parallel pass adjacent to the first. Similar to serpentine but with more rigid lane spacing. Best when the space has a clear long axis.
Perimeter-Plus-Interior
Walk the perimeter of the room first, then make one or two passes through the interior to capture the center. This works well for rooms with significant central features (large equipment, displays, fixtures) that need to be circled.
Side Shuffle (Crab-Walk)
Walk sideways through narrow corridors and tight spaces. This keeps both walls in the LiDAR field of view simultaneously. Movement is slower than a forward walk but produces complete dual-wall coverage in a single pass.
Real-time loop closure provides positive feedback. The L2 Pro and K2 both perform SLAM optimization during the scan. Watch the LixelGO preview as you complete a loop: visible tightening of the point cloud confirms loop closure detection. If the preview does not tighten when you expect it to, the loop did not register and you may need to walk the loop again at a closer return angle.
Long Corridors and Featureless Runs
Corridors accumulate drift faster than open spaces because there is less geometric variation for SLAM to correlate against. The repetitive geometry of long corridor walls offers fewer distinguishing features per unit of travel, so the system has less data with which to compute its position confidently.
Branching Frequency
- For corridors over 160 ft (50 m) without RTK: Branch into at least one room and return before reaching the far end. A straight out-and-back walk without branching accumulates uncorrected drift
- For corridors with RTK Fixed status: Length tolerance extends to approximately 500 ft (150 m) because absolute position is bounded by the RTK coordinate stream. Branching is still recommended but not as critical
- The branching room does not need to be on the route plan. Even a brief detour into any accessible space (a janitor's closet, an electrical room, a break room) creates a loop closure that resets accumulated drift
Side Shuffle in Narrow Corridors
For corridors under 6 ft (2 m) wide, the side shuffle technique gives SLAM more useful data than a forward walk. Both walls remain in the LiDAR field of view simultaneously, which provides the geometric variation that a forward walk in the same corridor cannot.
The side shuffle is slower than walking forward. Budget approximately 50% more time for any corridor walked sideways versus walked forward.
Targets in Featureless Corridors
For corridors over 200 ft (60 m) with no distinguishing features along their length, consider placing high-contrast SLAM anchor targets every 100 to 150 ft (30 to 45 m). These targets do not require surveyed coordinates to be useful: they give the SLAM system explicit high-contrast features to match against when the surrounding geometry is uniform. See Module 4: Positioning for target placement detail.
Multi-Floor Coverage
Buildings with multiple floors introduce additional planning considerations. Each floor is scanned in sequence, and stairwells or elevators provide the geometric continuity between them.
Order of Floors
- Start at the ground floor and work upward. The ground floor typically connects to the most exterior coordinate references (GCPs, RTK signal entry points, the building entry). Establishing the ground floor first anchors all subsequent floors
- Scan each floor as a continuous session from the moment you exit the stairwell or elevator on that floor to the moment you re-enter on the way to the next floor
- Do not stop the scan between floors. The stairwell or elevator is part of the same continuous session. Stopping and restarting between floors creates separate sessions that must later be joined with Map Fusion, which is more complex and produces lower alignment quality than a single continuous scan
Stairwell Coverage
See 3.3 Transitions for stairwell scanning detail. Stairwells take longer than their floor footprint suggests and are a high-failure transition. Plan time accordingly.
Elevator Coverage
Elevators present a special challenge: the elevator car itself moves vertically through space the scanner has never seen, with no visual or LiDAR continuity between floors during the ride. The SLAM system effectively loses track between the moment you enter the elevator and the moment you exit on the next floor.
- Do not scan inside elevator cars. Stop the scan before entering the elevator, ride to the next floor without scanning, and start a new scan session on the next floor
- Use Map Fusion in processing to join sessions separated by elevator rides. Each session must have sufficient overlap geometry with neighboring sessions for Map Fusion to align them. Where possible, use the stairwell rather than the elevator for at least one transition between any two floors, to provide alignment data
- Plan elevator and stairwell sequencing so every floor has at least one stairwell connection in the scan dataset. If a building has multiple stairwells and elevators, choose one stairwell as the canonical inter-floor connector for the scan
Three-Pass Capture Strategy for Interior Spaces
For maximum vertical coverage in interior spaces, especially for 3DGS output, use three passes at different heights within a single continuous session. This is the recommended approach for any high-deliverable indoor scan.
- Pass 1, chest height (approximately 5 ft / 1.5 m): Primary SLAM loop and core wall geometry. This is the foundational pass. Walk the full route at chest height
- Pass 2, overhead (approximately 7 to 8 ft / 2.1 to 2.4 m): Overhead views, ceiling lines, light fixture clarity, upper wall detail. Hold the device above head height or use the optional extension pole if available
- Pass 3, low (approximately 2 to 3 ft / 0.6 to 0.9 m): Baseboards, floor textures, under-furniture detail. Hold the device at hip height or kneel briefly at specific features
Maintain consistent height within each pass for even quality. Change heights gradually as you transition between passes, not abruptly when starting a new pass. The three passes must be a single continuous session, not three separate scans. Stopping between passes creates artifacts at the join points.
Three-pass scans take approximately 2.5 to 3 times as long as a single chest-height scan of the same area. The improvement in 3DGS quality is significant for visually demanding deliverables.
The Continuous Session Principle
SLAM performance depends on continuous trajectory data. Every time you stop the scan and start a new one, you introduce a discontinuity that processing must compensate for through Map Fusion. Map Fusion is a powerful tool, but it is not free: it requires additional setup, additional processing time, and produces lower alignment quality than a continuous scan.
When to Use a Single Continuous Session
- Single-floor projects: always one session
- Multi-floor projects: one session per floor, joined with Map Fusion at stairwells
- Projects under the RAM limit: Where the total scan duration fits within your processing machine's RAM capacity, use one session
When to Break Into Multiple Sessions
- Projects exceeding RAM limits: If the total scan duration would exceed your machine's processing capacity, split into multiple sessions of manageable duration and use Map Fusion to join them
- Elevator-separated floors: Each floor is a separate session, joined by stairwell traversal sessions or by Map Fusion alignment
- Multi-day projects: Each day's work is a separate session by definition. Plan day-to-day overlap to give Map Fusion sufficient geometric correspondence
- Operator fatigue: If a session would exceed 2 hours of continuous scanning, consider breaking it into multiple sessions with planned breaks between them. Fatigue-degraded posture in hour 3 produces less recoverable data than a clean second session
For complete Map Fusion planning detail, see Module 5: Map Fusion Fundamentals.
©2026 Alpine Reality Capture LLC • XGRIDS Pro Guide™ • Site Disclaimer

