A leading baseball team utilises semi-autonomous mobile robotic platform to scan their clubhouse and build a 3D model to give fans a first-hand experience of the interiors of the building
As a Major League Baseball team, the San Diego Padres often has to balance between requests from ardent fans interested in visiting their facilities, and protecting the privacy of the players. Even when visits are allowed, the franchise has to carefully guard the clubhouse — an almost sacred place for baseball followers — as most players have their personal stuff in the lockers. This is why the visits are limited.
The Padres team management approached the professionals of Sanborn Platform for Indoor Mapping (SPIN), an autonomous indoor mapping robot, for scanning their clubhouse. The team wanted to scan the clubhouse using SPIN and model it in 3D, so that fans could virtually walk through it from the comfort of their homes. An engineering grade accurate LiDAR point cloud could also help the facility managers run simulations in the future about alternate room layouts, acoustic modeling of the interiors and documenting the as-built conditions.
Once on-site, the SPIN robot took about 7 minutes to autonomously cover the entire clubhouse (about 3,700 sq feet in area). All along, it self-navigated between the furniture without requiring any pre-programmed trajectory or targets and avoided automatically obstacles (including the people present) in real-time. The clubhouse was measured by its onboard LiDAR sensors and an accurate 3D point cloud of laser points was produced in less than 20 minutes!
Point Cloud of the San Diego Padres’ Clubhouse, collected and generated by SPIN (acquisition time: 7 minutes; processing time: 18 minutes)
Traditional vs SPIN robot
Such a speedy workflow for accurately mapping and documenting as-built indoor environments has been unheard of till recently. In the current approach for scanning building interiors — by using terrestrial scanners on tripods — surveyors have to set up the equipments repeatedly to collect scans around occluded objects. This data acquisition process becomes cumbersome quickly in residential or office space settings where furniture and structural elements, such as walls, obstruct the laser scanner’s field of view. The post-processing also requires importing the scans one-byone (about 7 minutes per scan import) until the entire point cloud is stitched together, with or without targets added in the scene.
Sanborn has a rich history of accurately mapping the outdoor environments from both airborne and ground-based sensor platforms. Extending the mapping services portfolio to indoor environments is a natural progression of the firm’s spatial modeling capabilities. However, mobile mapping solutions for indoor environments have suffered from the lack of GPS, as it makes tracking the LiDAR sensors indoors challenging. Areas with bad or no GPS signal outside are usually mapped by interpolating the speed of the collection platform and/or measurements from the inertial measurement unit (IMU); this approach cannot be easily adopted for interior environments.
Recent advances made in the fields of robotics and computer vision have made it possible for tracking the sensors even in GPS-denied areas where robots need to autonomously navigate unknown environments. These capabilities are important in search and rescue (S&R) situations after natural disasters and for battlefield surveillance. Such robots scan the surrounding environment to build situational awareness, detect any openings (doors/windows) to explore and keep track of their position in space from the local point of origin.
Automatically mapping interiors
Sanborn researchers realised that if they could adapt the 3D modelling capabilities of S&R robots, the indoor mapping problem would be solved near-perfectly. A constant focus on accuracy resulted in the following functional specifications for Sanborn’s indoor mapping solution:
- Range measurements from 0.1m (minimum).
- Angular resolution of less than 0.5 degree.
- Scalable automated process for data acquisition with minimal human involvement.
- Robust enough to withstand extended use (3-4 hours at a time) in normal indoor environments.
- Efficient automated post-processing of data.
- Able to produce 3D point clouds of the environment and 2D floor layouts.
- Flexible enough for future sensor needs.
SPIN is a semi-autonomous mobile robotic platform designed for indoor mapping applications
Sanborn’s solution for indoor mapping is different from the existing approaches owing to its emphasis on providing an autonomous data acquisition platform, an affordable, scalable and repeatable solution, and an engineering grade accuracy point cloud. The latter is key since engineering grade accuracy enhances the usability of data beyond a purely visualisation perspective, i.e., it can act as the basis for solving and meeting real-world analytical problems.
The SPIN combines 2D Simultaneous Localisation and Mapping (SLAM) system based on the integration of laser scans in a planar map (using scan-matching) and an integrated 3D navigation system based on an IMU. The combination of high update rate simultaneous on-board 2D mapping, 6 degrees-of-freedom pose estimation and 3D modeling, consumes low computational resources and thus can be used on low-weight, low-power robots, and requires low-cost processors. The system also uses a depth camera to detect obstacles in front.
SPIN currently outputs the following data:  a 2D map of the indoor environment;  a 3D LiDAR intensity point cloud of the building interiors; and,  a topological map of the indoor elements. The dense and accurate 3D LiDAR intensity point cloud can be vectorised to extract the objects in the scene. This can further be used to build an inventory of the quantity, locations and dimensions of objects, including office furniture, computer/machine equipment, etc.
Sanborn’s indoor mapping initiative is aimed at scaling up the accurate documentation of as-built conditions. This is essential to allow the facility stakeholders (owners, engineers, architects and facility managers) to remodel the space, simulate alternate layouts and even update the floor plans — a common problem for old buildings.
In the near future, it will be commonplace to witness multiple self-navigating scanning robots quietly moving around different floors of multi-storied buildings simultaneously, going inside cubicles and office spaces, detecting and avoiding obstacles by themselves and returning to the starting point when they finish scanning. The possibilities of such advances for applications such as navigation, 3D printing and augmented reality video gaming are only limited by our imagination!
Sharad V. Oberoi,
Technical Development Manager,
The Sanborn Map Company, Inc.,
John R. Copple
The authors would like to thank the San Diego Padres management for their kind permission to scan and model the Clubhouse; Jeff Specht and Jason Caldwell for their help in arranging the site visit in San Diego; and LeaTrisha Bonati, Ryan Cutter and Tom Pendergrass for the 3D modeling from the LiDAR point cloud collected by the SPIN robot.