Weekly Robotics Newsletter Archive [7/16]
1) Stanford Engineers Design a Robotic Gripper for Cleaning Up Space Debris (2017).
INFO: This YouTube video shows a gecko inspired gripper created by Stanford researchers that in the future could be used for object manipulation in space and cleaning space debris.
4) Event Camera Helps Drone Dodge Thrown Objects.
INFO: Researchers from Robotics and Perception Group presented dynamic obstacle avoidance on a drone allowing it to dodge a ball thrown at it at 10m/s. The drone showcased in the article uses an event camera for localization and obstacle detection and in some experiments a motion capture system is used for getting ground truth information for both a drone and the ball.
5) This Two-Wheeled RC Car Is Rather Quick.
INFO: Compared to Bobble-Bot that we covered in the previous issue this self-balancing platform is a simpler build when it comes to software but we still found it quite interesting and entertaining to watch. The platform is based on the oDrive motor controller, a Teensy 3.6 micro computer and MPU6050 IMU. According to the captions in the attached video the platform is capable of driving at 40 MPH (64 km/h).
6) Underground Robots: How Robotics Is Changing the Mining Industry.
INFO: This article shows couple of use cases how robotics can be used in the mining industry for searching for rare metals and free humans from needing to be exposed to danger in mines. Part of the article focuses on the UX-1 robot developed as a part of the UNEXMiN project. UX-1 is an underwater robot with expected weight of 112 kg (247 lbs) and 0.6 m (2 ft) diameter and capable of operating at a maximum depth of 500 m.
7) Publication of the Week - Astrobee Robot Software: Enabling Mobile Autonomy on the ISS (PDF) (2018).
INFO: We mentioned Astrobee back in Weekly Robotics #35. The above paper presents the software architecture of Astrobee. Each Astrobee runs on 3 interconnected ARM processors communicating over ethernet and interfacing with 7 microcontrollers, 6 cameras, propulsion system etc. The system is composed of about 46 ROS nodelets. The low and medium level processors run Ubuntu 16.04, while the high level one runs Android (Nougat 7.1) that is used for guest science applications developed by research partners. The system does not require real-time kernel extensions due to the control loop running at low speed of 62.5Hz. The system has 3 sources of localization depending on the localization mode: 120° camera, AR markers for docking or depth camera for perching. Apart from the publication we found this talk by Andrew Symington to be a good summary of the software stack.
1) Humble Book Bundle: Artificial Intelligence & Deep Learning by Packt.
INFO: Pay what you want for those 25 books on AI & Deep Learning. By using the above link you can choose to support WeeklyRobotics and help us grow.
1) Aeolus Robotics (Various Countries and Remote) - Various Positions.
INFO: Founded to bring the first generation of multi-purpose robot assistants into service across the globe, Aeolus Robotics is a global company with offices in Taiwan, Poland, Austria and the USA. Integrating world-class, world-wide AI and Robot Systems genius with Taiwanese manufacturing aptitude, we are breaking new ground in general-purpose commercial-consumer robotics with capabilities in unstructured “human” spaces.
2) Magazino (Munich, Germany) - Robot QA Engineer.
INFO: Magazino develops and builds intelligent, mobile robots for intralogistics.
3) Cobalt Robotics (San Mateo, CA, US) - Robotics Software Engineer.
INFO: Cobalt builds autonomous security robots that keep buildings safe by combining the reliability of machines and the friendly face of human-in-the-loop “Robot Specialists.”
1) UNEXMiN Final Conference - 26 September 2019.
INFO: We mentioned UNEXMiN and UX-1 robot in 6th entry of this newsletter. On 26th of September this year the project results will be presented in Belgium and you can attend the event free of charge.
2) AI in Finance Summit, New York.
INFO: Discover advances in AI & machine learning tools and techniques from the world’s leading innovators across industry, research and the financial sector. As a media partner of the summit, Weekly Robotics readers can now get up to 40% off summit passes using discount code WR20. See the full agenda, list of speakers and registration details here. Summit passes to the Deep Learning Summit will also grant you access to the AI Assistant Summit and AI in Retail and Advertising Summit.
3) Deep Learning Summit, Europe.
INFO: Bridging the gap between the latest technological research advancements and real-world applications in business and society. Weekly Robotics readers can now get up to 40% off summit passes using discount code WR20. See the full agenda, list of speakers and registration details here. Summit passes to the Deep Learning Summit will also grant you access to the AI Assistant Summit and AI in Retail and Advertising Summit.
0) Anki Addresses Shutdown.
INFO: Back in the issue 37 we mentioned Anki closing down. The Anki addressed the shutdown in a pop up window on their website linked above. According to the statement Anki has contracted most senior leaders and engineers to provide a long term support for the devices. We will update you on any new information we find about Anki shut down.
1) A Comprehensive List of 3D Sensors Commonly Leveraged in ROS Development.
INFO: This ROS-Industrial page contains a list of affordable ROS-compatible 3D sensors. If you are looking for 3D sensors for your project this page might be a good first step.
2) Microsoft Launches a New Platform for Building Autonomous Robots.
INFO: At Build developer conference Microsoft announced an Azure-based platform that is said to provide AI solutions for automation tasks. You can find the project website here.
3) Digital Fabrication of Soft Actuated Objects by Machine Knitting 2019.
Morphing Matter Lab
INFO: This work done by Researchers from Carnegie Mellon University explores computationally-controlled machine knitting. By integrating tendons into the design the authors are able to create actuated soft objects. We recommend watching the two videos in the linked page and if you would like to know more - the page also has a link to the publication.
INFO: Bobble-Bot is a self balancing robot built with Raspberry Pi 3B, brushless DC motors and is running ROS (according to the article the authors are eyeing ROS2 implementation too). We enjoyed the comparison of non-real time and real time control that can be found in this log entry. You can find the project repository on GitHub. The project comes with a Gazebo model, allowing you to simulate the robot yourself.
5) The UZH-FPV Drone Racing Dataset.
Robot Perception Group
INFO: The above dataset provides camera images, IMU data, ground truth from a laser tracker, event data from an event camera and high resolution images from the pilot’s FPV camera. The goal of this dataset is to help advance the state of the art in high speed state estimation. At the moment the data is only available as a rosbag (a recording of ROS messages).
6) This Robot Hummingbird Is Almost as Agile as the Real Thing.
INFO: Researchers from Purdue University’s Bio-Robotics Lab have developed a robotic hummingbird. The robot weights only 12g and has a size of an actual hummingbird. In the videos you can see the robot being tethered for both control and power but according to the interview with Professor Xinyan Deng the robot should fly untethered in the future.
7) Publication of the Week - Sweeper: A Sweet-pepper Harvesting Robot, a Webinar (2019).
INFO: This 56 minutes webinar talks about a sweet pepper harvesting robot destined for indoor pepper farms and using heat pipes as rails. The first part of the webinar focuses on the project scope and result while the second part covers deep learning for pepper detection. The robot developed for this project is equipped with a 6 DOF robot arm mounted on a scissor lift and uses deep learning ready GPUs and the control system is based on ROS. The end effector used for harvesting is equipped with RGB + 3D ToF camera, an LED flashlight, a cutting tool and a catching device. The vision systems challenges a project like this faces are leaf occlusion, varying light conditions, ripe detection. The best results obtained during testing was 81% success rate (compared to ~100% required) and a cycle time of 24s to harvest a single pepper (compared to 10s required). To learn more about the project visit Sweeper website.
1) One Year of Working as a Robotics Consultant.
INFO: In this blog post Mat, the author of Weekly Robotics, sums up his first year as a robotics consultant. In the post, among other things, Mat describes his experience with UpWork and lists the projects he has worked on. The posts ends with potentially useful tips on helpful tools to use with ROS.
1) Sevensense (Zurich, Switzerland) - Various Positions.
INFO: Sevensense is committed to helping industries achieve new levels of efficiency by pushing the frontiers of mobile robotics. Our team in Zurich develops and delivers both software and hardware, providing a complete solution to the robot navigation problem.
2) wheel.me (Oslo, Norway) - Computer Vision Engineer.
INFO: wheel.me is a Norwegian IoT company with global ambitions, located in Oslo. We help people improve the way they live and work by enabling everything indoors to move effortlessly on smart wheels.
3) FarmWise (San Francisco, CA, US) - Various Positions.
INFO: At FarmWise, we build autonomous robots for agriculture. We work with and for growers to help them move away from chemical-based processes and face the labor crisis that is transforming the face of agriculture.
1) ROS Industrial (Melodic) Training Exercises.
INFO: This ReadTheDocs page from ROS Industrial contains what we would classify as intermediate to advanced exercises for ROS. Each set of exercises is accompanied by slides prepared by Southwest Research Institute.
2) The Hundred-Tonne Robots That Help Keep New Zealand Running.
INFO: In this video Tom Scott presents the 70 tonnes autonomous robots for moving ship containers that are being tested in New Zealand. According to the video the robots can stack containers with accuracy of couple of centimeters (which in itself can cause other problems as you can learn from the video). Quite interesting development is the local positioning system used by these robots.
3) Successful Demonstration of Breakthrough Blown-air Flight Technologies to Revolutionise Future Aircraft Design.
INFO: MAGMA is an UAV designed and developed by researchers at The University of Manchester in collaboration with engineers from BAE Systems. The article presents two falp-free fixed wing aircraft control methods. The demonstrators were: Wing Circulation Control: taking air from the aircraft engine and blowing it supersonically through narrow slots around a specially shaped wing tailing edge in order to control the aircraft and Fluidic Thrust Vectoring: controlling the aircraft by blowing air jets inside the nozzle to deflect the exhaust jet and generate a control force.
4) Seven Dreamers / Anki closes down.
The Robot Report/IEEE
INFO: The past week has been filled with bad news for those waiting for a rise of consumer robots. Seven Dreamers, a company working on the Laundroid laundry-folding robot close down. We were very surprised to learn that Anki (which we mentioned when we showcased vector_ros project one week ago) is also closing down due to failing to secure a new round of financing. The linked IEEE article raises a good point on what happens to customer’s robots if the cloud servers go down. This guest article by Guy Hoffman is an interesting analysis of failed social robot startups.
5) The Best Robots on Four Legs with Marc Raibert (Boston Dynamics).
INFO: In this video from TechCrunch Marc Raibert, CEO of Boston Dynamics, discusses the current state of Boston Dynamics and offers some interesting insights on development of robotics products. At 7:30 you can see how components of Spot Mini and the robots themselves are tested.
6) Hitachi Establishes the Corporate Venture Capital Fund to Support Startups as an Innovation Partner.
INFO: Hitachi is establishing Hitachi Ventures GmbH that will start on 1st of June and will start with $150M fund.
7) Publication of the week - NASA Investigative Summary: TaurusXLT8 and T9 Mission Failures (2019)(PDF).
INFO: Sapa Profiles, Inc. (SPI) provided NASA with faulty materials that ended up being a root cause of 2 launch failures and causing a loss of $700 million. The PDF file linked above is a summary of the investigation carried out by NASA.
1) Sense Photonics (Edinurgh, UK / Research Triangle, San Francisco US) - Various Positions.
INFO: Sense Photonics produces advanced LiDAR and 3D sensor solutions for the autonomous vehicle, UAV and industrial automation markets.
2) Enway (Berlin, Germany / Singapore, Singapore) - Various Positions.
INFO: We are building the technology for high precision autonomous service vehicles. Enabling cleaner and more liveable cities.
1) Laying the Ground for Robotic Strategies in Environmental Protection.
INFO: Researchers at the Wyss Institute have developed Romu, a robot designed to drive interlocking sheet piles into the ground to help stabilize soil. The robot is using linear actuators and a vibratory hammer to press the sheets into the ground.
2) Snake-inspired Robot Uses Kirigami for Swifter Slithering.
The Robot Report
INFO: Kirigami is a papercraft that uses cuts to change the properties of the material. By carefully choosing cuts in a shell the Scientists can ‘program’ the skin to deform in a desired sequence.
3) NASA’s Robonaut to Return to Space Station With Legs Attached.
INFO: Robonaut is currently being prepared to go back to ISS, this time with legs attached, that are key to the Robonaut Project research objectives. The above article contains an interview with Julia Badger who is the Robonaut Project Manager, we highly recommend going through it! If you happen to use ROS and Gazebo then you might be able to run the Robonaut 2 simulation by following this wiki.
4) Paris Firefighters Used This Remote-Controlled Robot to Extinguish the Notre Dame Blaze.
INFO: In the previous issue of Weekly Robotics we’ve mentioned the robots used during the Notre Dame fire. This articles from IEEE provides more details about the Colossus, a 500 kg (1100 lbs) firefighting robot. We were quite surprised to learn that the robot is capable of moving 1,000 kg (2200 lbs) of payload.
5) ROS Package for Anki Vector Home Robot.
INFO: Omri Ben-Bassat has created an unofficial ROS package for Anki Vector, a small wheeled home robot. The package is open source and is a wrapper around Vector SDK. It provides interfaces for the camera, velocity commands, animations, saying text etc.
6) Velodyne Lidar Partners with Nikon to Mass-produce Cheaper Lidar Sensors.
INFO: “Sendai Nikon Corporation, a Nikon subsidiary, will manufacture lidar sensors for Velodyne with plans to start mass production in the second half of 2019”. Hopefully the ramp up in production will result in cheaper multi-axis LiDARs on the market.
7) Publication of the week - Pseudo-LiDAR from Visual Depth Estimation:Bridging the Gap in 3D Object Detection for Autonomous Driving (2019).
INFO: We found the above paper cited in several articles (, ) that surfaced after Elon Musk made a statement about LiDARs being a fool’s errand during Tesla autonomy days earlier this week. We’ve decided to see ourselves what the paper says and if stereo cameras will beat LiDARs any time soon. The paper has been written by Cornell University Researchers. In the paper they propose an alternative method for representation of 3D information obtained from stereo cameras so that it resembles LiDAR data (pseudo-LiDAR) and therefore LiDAR based object detection algorithms can be applied to it. The experiment setup is based on the KITTI benchmark and there are two algorithms that Researchers evaluate: F-POINTNET-v1 and AVOD-FPN. Looking at the results table the chosen approach is a significant improvement over other camera based methods however at high Intersection over Union there is still a high gap between stereo based method described and Lidar + monocular camera based method. In case of Bird Eye View representation with AVOD algorithms and the IoU of 0.7 the researchers obtained a result of 74.9 (89.4 for LiDAR + mono camera) in Easy category and 49.0 (79.3 for LiDAR + mono camera) in Hard category.
1) Humble Book Bundle: Electronics + 3D Printing by Make.
INFO: This Humble Book bundle features 24 books from Make that are worth $401. In the bundle you will find books related to 3D printing, electronics, programming and even drones. If you purchase the bundle using the above affiliate link you can choose to support Weekly Robotics.
1) Moley (London, UK) - Various Positions.
INFO: Moley has created the world’s first robotic kitchen. Featuring an advanced, fully functional robot integrated into a beautifully designed, professional kitchen, it cooks with the skill and flair of a master chef. The prototype was premiered to widespread acclaim at Hanover Messe, the international robotics show.
2) Sonnet.ai (Seoul / Daegu, South Korea) - Senior Software Engineer, Autonomous Driving.
INFO: Sonnet.ai is a tech startup bringing artificial intelligence to autonomous vehicles and precision medicine. We founded sonnet.ai because we believe that artificial intelligence technology has the potential to save lives and transform our societies, and we think this is the right team to do it.
3) Barrios Technology (Houston, TX, US) - Software Developer.
INFO: Barrios Technology Ltd. is a woman-owned and operated small business headquartered in Houston, Texas. We provide a full spectrum of engineering, operations and related technology services in support of the aerospace community and have successfully supported NASA for 38 years.