Exploring the Future of Elephant Robotics

Elephant Robotics builds compact collaborative arms, teleoperation hardware, and classroom kits, and its current lineup shows where small-scale robotics is heading: toward robots that learn tasks from human demonstration and then carry them out on their own. If you are weighing one of its arms or platforms for a lab, a classroom, or a home workshop, this article covers what the brand is doing now and which product is the right starting point.

Where is Elephant Robotics headed?

Elephant Robotics was founded in 2016 in Shenzhen and is best known for the myCobot family, the compact six-axis collaborative arms that put real kinematics on classroom desks. The company reports more than 10,000 myCobots sold across 50-plus countries. That base gave it the scale to move into a workflow that used to belong only to well-funded labs.

That workflow is imitation learning: a person demonstrates a task by moving a leader arm, the motion is recorded, and a follower arm learns to reproduce it. The newest hardware, the myArm C650, myArm M750, and myController S570, exists to make collecting that demonstration data affordable. The Elephant Robotics brand page covers the company background in more detail.

The second theme is consistency. Python and ROS run across the whole catalog, URDF models (the robot description files that simulators read) are published for the major simulators, and the same SDK concepts carry from a desktop arm up to the LIMO Cobot mobile base. A program that starts with tabletop pick-and-place can add mobility later without changing ecosystems.

How does the teleoperation pair work?

The myArm C650 and myArm M750 are two halves of one system. Both are 6+1 DoF arms (DoF means degrees of freedom, the number of independently controlled joints). The C650 is the leader: you move it by hand and it streams endpoint coordinates or joint angles at 50 Hz through a fingertip controller with two smart buttons. It carries no payload and has no gripper; it is a data collector, not a manipulator.

The myArm M750 is the execution side. It has a 750 mm reach, a 500 g rated payload (1 kg peak), plus or minus 1 mm repeatability, and an included 1-DoF parallel gripper. Pair it with a C650 for leader-follower training, or run it on its own through Python or ROS for pick-and-place and vision projects.

Elephant documents ROS 1 (Kinetic, Melodic, and Noetic) and ROS 2 (Foxy) for the pair, with MoveIt motion planning and RViz visualization packages. Isaac Sim, MuJoCo, and PyBullet model files are not published, but because the C650 streams joint data through the Python SDK, it can drive a simulated arm through a small bridge script.

What does the myController S570 add?

The myController S570 scales the same idea to two arms. It is a wearable exoskeleton that fixes onto a vest and tracks both of the operator's arms: 12 DoF in total, read by 4096-count magnetic encoders sampling at up to 100 Hz. Each arm section has a 570 mm reach and the whole controller weighs 1.5 kg (3.3 lb). It connects over USB Type-C or wirelessly over Wi-Fi or Bluetooth, one wireless service at a time.

Officially supported pairings include the myCobot 280 and 320, myCobot Pro 630, mechArm 270, myArm M750, myBuddy 280, myAGV, the Mercury humanoid, and Universal Robots arms. ROS 1 and ROS 2 publishers expose the joint angles, so any robot that subscribes to standard joint-state topics can follow the operator. The exoskeleton is passive: it has no motors and only senses the wearer's motion.

Why does mobile manipulation come next?

A desktop arm can only reach what sits in front of it. The LIMO Cobot answers that limit by mounting a myCobot 280 arm on an AgileX LIMO PRO wheeled base. The base carries an EAI T-mini Pro LiDAR (a spinning laser rangefinder used for mapping), an Orbbec Dabai depth camera, and an NVIDIA Jetson Orin Nano running Ubuntu 20.04 with ROS 1 and ROS 2 Foxy guides.

SLAM (simultaneous localization and mapping) through Cartographer and Nav2-based navigation are provided, and MoveIt handles the arm's grasp planning. The standard workflow is to map a room once by tele-operating the base, then send goal poses; the base drives to the target and stops before the arm executes the grasp. The arm keeps its 250 g payload, and the base is rated for a 4 kg load in total, which includes the 850 g arm.

AgileX rates the base at about 2.5 hours of working time and 4 hours of standby per charge from its removable 12 V, 10 Ah battery, with about 2.5 hours to recharge. It is built for flat indoor floors; the arm and sensors are not weatherproof.

What about classrooms that need a turnkey kit?

Not every program wants to assemble a research pipeline. Elephant Robotics also builds three institutional kits that RoboticsSelect handles through its quote process:

• The UltraArm Conveyor Belt Kit pairs an ultraArm P340 arm (340 mm working radius, 650 g payload) with an 800 mm by 100 mm motorized conveyor. Open-source Python programs for color recognition and QR code sorting run out of the box, then become the baseline for student projects.
• The Compound Robot Intelligent Logistics Kit is a warehouse-automation teaching cell: a myAGV mobile platform with an arm mounted eye-in-hand, a conveyor, and a vision system for OCR, ArUco, and QR code detection, plus a college-level course with 10 experimental cases covering SLAM mapping, path planning, item detection, loading, and transport.
• The Portable AI Educational WorkStation packs compute, a full-size keyboard, a mouse, and a joystick into a 600 by 400 by 250 mm unit that weighs 11 kg without the arm. It boots into Ubuntu with ROS 1 and ROS 2, supports one-click startup and drag-and-teach, and teaches QR code recognition, image segmentation, feature extraction, and YOLOv5 object detection with the included colored blocks and cards. It is built for K-12 and early college.

All of these sit in the robotic kits collection, and the robotic kits buying guide explains how to match a kit to grade level and budget.

How should a beginner choose?

Choose by the outcome you need, not by the most advanced spec sheet:

1. Learning imitation learning or teleoperation: start with a C650 and an M750. Most users record 20 to 100 demonstrations per task and add a wrist camera for vision-conditioned policies.
2. Bimanual research: use the S570 exoskeleton as the leader and drive two execution arms.
3. Mobile manipulation or a fetch-and-carry course: the LIMO Cobot arrives with SLAM, Nav2, and MoveIt already integrated.
4. A classroom starting from zero: the Portable AI Educational WorkStation for K-12, or the UltraArm Conveyor Belt Kit for automation and sorting lessons.
5. A makerspace or advanced hobbyist: one M750 with Python, plus a USB camera on the optional camera flange for OpenCV projects.

What does buying through RoboticsSelect include?

RoboticsSelect is an authorized Elephant Robotics dealer, so every arm and kit ships through the manufacturer's official channel and the full 365-day warranty on mechanical structure, power supply, and control systems applies from delivery. Accessories such as servo grippers, vacuum pumps, sensors, and camera modules follow the shorter component terms in Elephant's warranty policy.

Shipping is free to the contiguous 48 U.S. states, and U.S.-based product-trained support is available at support@roboticsselect.com or (412) 415-7375. Education buyers can order by purchase order and get pre-sales configuration advice; the quote-based kits are handled the same way. A price match policy and a financing page are available, and the store's standard 30-day return window applies unless a product page states a different term.

The takeaway

The future of Elephant Robotics is a pipeline, not a single arm: collect demonstrations with a C650 or S570, execute with an M750, and extend to mobile manipulation with the LIMO Cobot, all on the same Python and ROS foundation. Start with the piece that matches your immediate goal and browse the full Elephant Robotics collection to compare the current lineup.