When PickNik Robotics sat down to design MoveIt Pro 9.0, the company made a bet that the robotics industry is about to learn an uncomfortable lesson: perfect autonomy is a fantasy, and reliable imperfection is a business.
PickNik, a Boulder-based robotics software company, released the ninth major version of its commercial MoveIt Pro platform this month. The release is the company's first significant update in nearly a year, and it represents a deliberate philosophical pivot as much as a technical one. The headline features include enhanced perception-to-motion capabilities that let robots scan a surface, generate a collision-aware tool path, and execute it across variable geometry. The teleoperation system has been rebuilt from the ground up. Underneath it all, PickNik has fully severed its remaining dependencies on the open-source MoveIt 2 project, replacing core algorithms with implementations it claims are 35 times faster for inverse kinematics solvers, four times faster for motion planning, and 30 times faster for Cartesian planners.
But the most revealing thing about the release is how CEO Dave Grant talks about it.
"The industry got overly focused on 100% autonomy somewhere along the way," Grant told The Robot Report. "Coming from outside the robotics industry, if I hired a new employee, would I expect that new employee to be 100% independent always, or would I be okay if nine times out of 10 they are independent? Once out of 10 times, they have to phone home for some help, or ask some questions."
That framing matters because it represents a growing split in how robotics companies are choosing to position themselves. The companies chasing full autonomy tend to point to demos. The companies building for deployment tend to talk about what happens when the demo ends. PickNik is clearly in the second camp.
What the robot actually does now
MoveIt Pro 9.0 centers on what PickNik calls a perception-to-motion pipeline. A robot scans an object using a depth sensor, builds a point cloud of its geometry, and then automatically generates rasterized Cartesian tool paths across the contours. The system handles the path planning, collision checking, and joint limit enforcement in a single workflow. The company positions it for surface-driven tasks: spraying, washing, sanding, grinding. In practice, that means applications like car washing, industrial cleaning, and surface coating where the geometry of each object varies enough that pre-programming a fixed path is impractical.
Denver-based Autowash is one of PickNik's early customers in the vehicle-care sector. The company uses MoveIt Pro to automate washing across varying vehicle shapes and sizes. According to Grant, Autowash cut energy and water use by around 40% because the system could generate car wash plans specific to each vehicle rather than running a generic cycle. "Every vehicle presents different contours and surface conditions," said Dennis Dreeszen, co-founder and CEO of Autowash. "MoveIt Pro's perception-to-motion pipeline allows us to generate consistent coverage while maintaining safe operation around unpredictable geometry."
Hivebotics, a Singapore-based developer of autonomous restroom cleaning robots, also uses MoveIt Pro to manage adaptive surface coverage across varied bathroom geometries. Hivebotics had originally attempted to build its own stack on top of open-source tools before concluding it would take too long and cost too much. "Speed to implementation and speed to production are the key metrics for them," Grant said. CleanBotix, a Grand Rapids company focused on food processing plant sanitation, is another customer using the platform for tasks that traditionally required manual intervention.
The teleoperation overhaul is the other significant piece of the release. MoveIt Pro 9.0 replaces the MoveIt Servo-based joint jogging system with a new Joint Velocity Controller that PickNik says offers smoother motion, built-in collision checking, and multi-arm support. The system also now accepts Meta Quest controllers for teleoperation data collection, allowing operators to record demonstration data that can be used to train end-to-end machine learning policies. This is explicitly designed for what PickNik calls "AI data collection" and "diffusion policy training" workflows, a bet that the future of manipulation involves learning from human operators rather than pure planning.
The NASA contract, and why it matters
On February 11, 2026, Motiv Space Systems announced a contractual agreement with PickNik to provide software for NASA's Fly Foundational Robotics (FFR) mission. FFR aims to demonstrate autonomous and ground-supervised robotic manipulation in low Earth orbit, supporting NASA's broader In-space Servicing, Assembly, and Manufacturing objectives. PickNik will deliver a flight runtime configuration of MoveIt Pro suitable for the mission compute environment, along with a ground-based operator terminal for mission planning and validation. The work will use Space ROS, an open-source distribution of ROS maintained by the Open Source Robotics Foundation with PickNik and NASA as primary contributors.
The NASA connection is not incidental. PickNik's teleoperation capabilities exist in part because NASA will not allow a robot to operate in space without a human in the loop. "It is part of our core DNA to have that teleoperation," Grant said. That constraint from spaceflight has become, in Grant's telling, a commercial feature rather than a limitation.
The business model argument
PickNik licenses MoveIt Pro commercially, with pricing tied to deployment scale. The comparison page for the platform makes the case explicitly: the open-source MoveIt 2 is appropriate for students and university researchers; MoveIt Pro is for commercial companies, research labs, and government agencies. The platform includes pre-built behaviors, a behavior tree editor, integrated simulation, and what PickNik describes as commercial-grade software with extensive unit testing and automotive-industry quality standards.
The company is also explicit about what it is not: it is not in the business of selling the dream of fully autonomous robots. "If you want to create a bad demo that will never see the light of day, you can do that faster than ever today," Grant said. "So we are really shifting our focus to the reliability of the platform, and executing with as close as possible to 100% reliability."
The question is whether that philosophy translates. The robotics industry has spent years marketing the autonomous future. Convincing customers to pay for a platform that explicitly acknowledges the limits of autonomy is a different kind of sell. PickNik is betting that the companies winning with robotics right now are the ones solving specific business problems reliably, not the ones building toward HAL 9000.
Sources: The Robot Report | PickNik March Newsletter | PickNik Release Notes | SpaceNews | PickNik MoveIt Pro vs MoveIt 2
† If the article intends to claim Meta Quest controller support, add verification language or confirm with source. If not claiming this specifically, no change needed.
†† Add source citation for Boulder location or remove the geographic descriptor.
††† Add independent source confirmation for Hivebotics' location, or remove the geographic descriptor.
† If the article intends to claim Meta Quest controller support, add verification language or confirm with source. If not claiming this specifically, no change needed.
†† Add source citation for Boulder location or remove the geographic descriptor.
††† Add independent source confirmation for Hivebotics' location, or remove the geographic descriptor.
