TECHNOLOGY PLATFORM
Motion Planning AI: Collision-Free Trajectories for Robotic Palletizing
Viroteq motion planning is the AI layer that turns a stacking decision into a safe, fast, collision-free robot trajectory. Sub-10ms path computation keeps cycle time tight, joint-limit and singularity optimisation keeps movement smooth, and payload-aware planning keeps every pick and place provably collision-free. As a result, the planner never becomes the bottleneck inside a Viroteq cell — instead, it unlocks the full speed of FANUC, ABB, KUKA, Universal Robots, Yaskawa, and Stäubli arms running in real production.
- ✓ Sub-10ms path computation per pick
- ✓ Collision-free trajectories, always
- ✓ Joint-limit and singularity optimisation
<10 ms
Path computation per pick
Brand-Agnostic
FANUC, ABB, KUKA, UR, Yaskawa
6 & 7-Axis
Multi-axis kinematics support
1000+
Deployed cells worldwide
Why Motion Planning Determines Cycle Time and Reliability
Motion planning is the difference between a robot that looks impressive in a demo video and a robot that earns its keep on a real production line. Generic path-planning libraries can find a route from A to B, but a production-grade trajectory engine must compute that path inside a tight cycle window, validate it against every joint limit, every singularity, and every obstacle in the cell, and stay deterministic across millions of repetitions per year.
Joint limits are the first hurdle. A six-axis robot has six hard mechanical stops, and a naïve planner happily plans toward configurations that hit them mid-cycle. Singularities are the second — robot poses where small Cartesian moves require huge joint velocities, which trip safety controllers and stall the line. Furthermore, collision avoidance is not just about the robot body. The gripper, the payload, and the dynamically changing pallet stack all have to be modelled, otherwise the planner produces trajectories that crash into freshly placed cases.
Payload-aware trajectories add another axis of complexity. A 25 kg case carried at full reach behaves very differently from an empty gripper at the same pose — accelerations, inertias, and safe approach angles all shift. Open-source libraries like MoveIt for ROS provide good academic baselines, but turning them into something safe enough to run unattended on a 24/7 line requires substantial engineering on top.
Viroteq’s technology platform closes that gap. The engine runs at the edge on Industrial PCs, validates against the full cell model on every cycle, and integrates with the safety architecture defined by industry-standard practices. Therefore, every trajectory is provably collision-free before the controller receives it, and the line runs at full robot speed without compromise.
How Motion Planning Works in Viroteq Cells
Every Viroteq planning cycle starts with a grasp pose produced by the StackrBrain runtime. The grasp pose describes where the gripper must approach the item, the goal pose describes where it must place the item on the pallet, and the cell model describes everything that must not be touched in between.
First, the planner runs a sampling-based path search through joint space using RRT-style and PRM-style algorithms, trained on production data to bias the sampler toward fast, smooth candidates. Next, an optimisation pass smooths the candidate path, removes redundant via-points, and respects joint velocity and acceleration limits. Then a final validation step checks the trajectory against every obstacle in the cell, the payload swept volume, and the safety zones defined for the cell.
Finally, the validated trajectory streams to the robot controller — FANUC, ABB, KUKA, Universal Robots, Yaskawa, or Stäubli — through native interfaces or the appropriate motion bridge. The whole pipeline runs in under 10 ms for typical six-axis arms, so it slots inside the cycle window of even the fastest palletizing automation stations.
Viroteq Products Powered by Motion Planning
Three Viroteq products rely on the same motion planning engine — RobotStackr OS for consistent palletizing runs, RobotStackr OTF for mixed-case real-time stacking, and RobotDepalr for inbound depalletizing. Therefore, one runtime drives every robot in the cell across palletizing, depalletizing, and mixed workflows.
RobotStackr OS
RobotStackr OS uses the trajectory engine to drive consistent single-SKU palletizing at full robot speed. Pre-validated patterns combined with optimised joint-space paths deliver maximum throughput without singularity stalls or collision events.
RobotStackr OTF
RobotStackr OTF re-plans every trajectory in under 10 ms as cases arrive in unknown sequence. Real-time replanning makes mixed-case palletizing as smooth as single-SKU, with no pre-programmed paths and no manual teach steps.
RobotDepalr
RobotDepalr uses the trajectory engine to clear inbound pallets layer by layer, replanning around tilted, shifted, or partially picked stacks. Therefore, every pick avoids neighbouring cases and the cell stays productive even when supplier pallets arrive imperfectly stacked.
Core Capabilities of Motion Planning AI
Collision Avoidance
Full kinematic and dynamic collision checks against robot body, gripper, payload, pallet stack, and safety zones. Every trajectory is provably collision-free before the controller receives it, which keeps the cell safe at full robot speed and prevents costly crash events on unattended shifts.
Joint-Limit Optimisation
Trajectories stay clear of hard joint stops and kinematic singularities through joint-space optimisation. The planner scores candidate paths on reach margin and smoothness, then rejects anything that would force costly mid-cycle reorientation. Therefore, the robot runs predictable, repeatable, full-speed cycles every time.
Payload-Aware Trajectories
The planner factors in payload mass, inertia, and swept volume on every cycle. As a result, it picks safe approach angles, respects acceleration limits when carrying heavy items, and validates clearance for tall or wide payloads. The same engine handles empty grippers and 30 kg cases without operator intervention.
What Is Motion Planning? Definition and Glossary
Motion planning is the algorithmic process by which a robot decides how to move from a start pose to a goal pose without hitting obstacles, exceeding joint limits, or entering singular configurations. In robotic palletizing, it converts a high-level instruction such as “pick this case and place it at coordinate X, Y, Z” into a fully validated joint-space trajectory that the robot controller can execute. Good path computation is fast, deterministic, and provably collision-free — which is why it sits at the heart of every robotic palletizing cell that runs reliably at full speed.
Furthermore, the topic matters because the alternative — manually programmed waypoints — does not scale to mixed-case workloads, dynamic pallet stacks, or changing item geometries. As a result, modern AI-powered path computation is what makes brand-agnostic robotic palletizing economically viable across every industry Viroteq serves.
High-Speed Production
FMCG bottling, canning, and packaging lines where the planner has to slot inside an aggressive cycle window. Sub-10ms path computation keeps every pick at full robot speed.
Mixed Contractor Orders
Building-materials and industrial distribution where each pallet contains a different mix. The planner replans every trajectory for each new SKU sequence without manual teach steps.
Cold-Chain Cells
Sub-zero environments where motion smoothness matters for gripper longevity and product integrity. The engine optimises acceleration profiles to avoid thermal-shock-related issues on cold cases.
Multi-Robot Cells
Dual-arm depalletizing or twin-cell palletizing where two robots share a workspace. The planner negotiates time-synchronised trajectories so the arms never collide and hand-offs stay seamless.
Outdoor Robots
Yard and outdoor depalletizing cells where wind, weather, and uneven ground change the cell model dynamically. The engine revalidates trajectories on every cycle to stay safe.
Heavy Payload Loads
Building-materials and industrial parts above 25 kg where payload-aware trajectories keep accelerations safe and approach angles respectful of gripper, robot, and worker safety.
High-Speed Palletizing
The trajectory engine sustains 1000+ cases per hour with sub-10ms path computation and joint-optimal moves on six-axis FANUC, ABB, and KUKA arms.
Multi-Robot Coordination
Time-synchronised trajectory coordination across two or more robots sharing a workspace, with conflict resolution and negotiated hand-off zones for seamless cycles.
Heavy Payload Cells
Payload-aware path computation for 25 to 50 kg items — accelerations, inertias, and approach angles are validated on every cycle for safe heavy lifting.
<10 ms
Trajectory computation per pick
1000+
Cells running Viroteq trajectory AI
99.9%
Trajectory validation reliability
Engineered for Real-World Production Constraints
A path-planning engine that works in an academic demo is not the same thing as one that survives a 24/7 production floor. Real cells have hard joint limits that vary by robot vendor, kinematic singularities that trip safety controllers, and payloads that change inertial behaviour every cycle. Viroteq’s planner treats all of these as first-class constraints rather than corner cases.
Joint-limit handling is built into the optimiser. Every candidate trajectory is scored on reach margin so the planner naturally prefers paths that stay clear of mechanical stops. Singularity avoidance uses the manipulability index — a measure of how far the robot is from a singular configuration — to steer the search away from poses that would force enormous joint velocities to track small Cartesian moves. Furthermore, payload effects feed into the dynamic check: a 30 kg case at full reach gets a different acceleration profile than the same robot moving empty.
Safety zones are integrated directly into the planning model. Light curtains, fence boundaries, and dynamic obstacles from vision and sensor inputs all become exclusion volumes that no candidate trajectory may enter. As a result, every plan is consistent with the safety architecture defined by industry-standard practices such as ISO 10218 robot safety standards, which guides how collaborative and industrial robots must behave around people.
Finally, the engine balances time-optimal versus energy-optimal trajectories. A time-optimal trajectory minimises cycle time and maximises throughput. An energy-optimal one minimises joint accelerations and reduces wear on the gripper, robot, and mechanical structure. Customers can tune the trade-off through configuration. Therefore, the same runtime serves a high-speed FMCG line that prioritises cases-per-hour and a heavy-payload industrial cell that prioritises long-term mechanical life — without forking the software stack.
Integrating Motion Planning With Your Robot Brand
The Viroteq planner is brand-agnostic by design. Native interfaces exist for FANUC, ABB, KUKA, Universal Robots, Yaskawa, and Stäubli, so the same runtime drives every robot in the cell without vendor SDKs, proprietary teach pendants, or middleware boxes. Communication runs over REST API and WebSocket, which keeps the integration surface small and reviewable for plant IT teams. Furthermore, the planner accepts standard Denavit-Hartenberg parameters as well as custom kinematic chains, so non-standard arms, gantries, and seven-axis robots are supported through configuration rather than scripting.
As a result, the trajectory engine travels with the customer rather than the robot vendor. Mixed-brand fleets across cells keep one operator workflow, one HMI, and one set of API endpoints. In addition, the runtime sits on Industrial PCs inside the cell, so path-computation latency is bounded at the controller rather than the cloud, and the line stays operational during external network outages or audits.
FANUC
Native trajectory bridge for FANUC R-30iB and R-30iA controllers via Stream Motion and PC interfaces.
ABB
Trajectory streaming for IRC5 and OmniCore controllers through Externally Guided Motion interfaces.
KUKA
KUKA KR C4 and KR C5 controllers driven via RSI for low-latency trajectory streaming.
Universal Robots
UR3, UR5, UR10, UR16 and UR20 cobots driven through the Real-Time Data Exchange for collaborative trajectory control.
Motion Planning AI — Frequently Asked Questions
See Motion Planning AI in Production
Book a personalised demo and watch sub-10ms collision-free trajectories drive your robot of choice — FANUC, ABB, KUKA, UR, Yaskawa, or Stäubli — in a real Viroteq palletizing cell.
AI Motion Planning Built for Real-World Robotics
Bring your robot brand, payload spec, and cell layout — Viroteq specialists will map a motion planning deployment path that fits inside your existing cell and keeps the line running.
