KORTHOS / ECOSYSTEM / PRODUCT

TALOS

About

About TALOS

TALOS is PAL Robotics’ torque-controlled humanoid with sensing at every joint, allowing researchers to study contact forces as well as commanded motion. Its arms can each carry six kilograms at full extension, supporting experiments with tools and heavy objects alongside bipedal balance and locomotion.

An EtherCAT network supports fast control loops, while ROS interfaces expose encoders, the inertial unit, joint torque readings and wrist and ankle force sensors. PAL supplies a dynamics model and an initial state estimate derived from inertial sensing and kinematics. Researchers can use these components with prioritised task-and-constraint controllers to coordinate the robot’s limbs and balance.

Capabilities

Source-backed capabilities

  • Operating environments: Research stair setup without prior terrain knowledge
  • Control modes: Autonomous
  • Stair climbing

    Research demonstration - 2023-12-30 - Climb 15 cm stairs without a prior terrain map and adapt when the stairs are moved. - environment: Research stair setup without prior terrain knowledge

    Source

Deployments

  • Customer Site

    1 TALOS humanoid at Waterloo RoboHub - 2018-11-07 - University of Waterloo RoboHub - research humanoid platform - PAL Robotics TALOS joined the University of Waterloo RoboHub as an experimental platform for control, planning, industrial automation and human-robot interaction research.

    Source

Research

  • Words2Contact: Identifying Support Contacts from Verbal Instructions Using Foundation Models

    2024-11-22 - Research Publication - This paper presents Words2Contact, a language-guided multi-contact placement pipeline leveraging large language models and vision language models. Our method is a key component for language-assisted teleoperation and human-robot cooperation, where human operators can instruct the robots where to place their support contacts before whole-body reaching or manipulation using natural language. Words2Contact transforms the verbal instructions of a human operator into contact placement predictions; it also deals with iterative corrections, until the human is satisfied with the contact location identified in the robot’s field of view. We benchmark state-of-the-art LLMs and VLMs for size and performance in contact prediction. We demonstrate the effectiveness of the iterative correction process, showing that users, even naive, quickly learn how to instruct the system to obtain accurate locations.

    Source
  • TALOS torque-controlled locomotion in unknown environments

    2023-12-30 - Research Publication - PAL and TOWARD describe an onboard TALOS locomotion system combining LiDAR mapping, contact planning, predictive control and inverse dynamics. Their physical demonstration climbs 15 cm stairs without a prior terrain map and adapts after stairs move.

    Source
  • Practical whole-body elasto-geometric calibration of a humanoid robot: Application to the TALOS robot

    2023-06-01 - Research Publication - The paper applies whole-body elasto-geometric calibration methods to the TALOS humanoid robot.

    Source
  • Benchmarking Whole‐Body Controllers on the TALOS Humanoid Robot

    2022-03-04 - Research Publication - This article presents a comparison of three control schemes applied on the commercially available TALOS humanoid robot.

    Source
  • Whole-body teleoperation of the Talos humanoid robot: preliminary results

    2021-05-30 - Research Publication - Maps an operator’s motion-capture measurements to Cartesian hand targets for the TALOS robot. Uses a task-based inverse-dynamics controller while accounting for balance, self-collision and the robot’s kinematic constraints. The authors’ 2021 activity report documents Xsens-suit teleoperation and links the associated whole-body-control implementation.

    Source
  • Variational Inference with Mixture Model Approximation for Applications in Robotics

    2020-05-01 - Research Publication - We propose to formulate the problem of representing a distribution of robot configurations (e.g. joint angles) as that of approximating a product of experts. Our approach uses variational inference, a popular method in Bayesian computation, which has several practical advantages over sampling-based techniques. To be able to represent complex and multimodal distributions of configurations, mixture models are used as approximate distribution. We show that the problem of approximating a distribution of robot configurations while satisfying multiple objectives arises in a wide range of problems in robotics, for which the properties of the proposed approach have relevant consequences. Several applications are discussed, including learning objectives from demonstration, planning, and warm-starting inverse kinematics problems. Simulated experiments are presented with a 7-DoF Panda arm and a 28-

    Source
  • Motion Planning with Multi-Contact and Visual Servoing on Humanoid Robots

    2020-01-01 - Research Publication - This paper describes the implementation of a canonical motion generation pipeline guided by vision for a TALOS humanoid robot. The proposed system is using a multicontact planner, a Differential Dynamic Programming (DDP) algorithm, and a stabilizer. The multicontact planner provides a set of contacts and dynamically consistent trajectories for the Center-Of-Mass (CoM) and the Center-Of-Pressure (CoP). It provides a structure to initialize a DDP algorithm which, in turn, provides a dynamically consistent trajectory for all the joints as it integrates all the dynamics of the robot, together with rigid contact models and the visual task. Tested on Gazebo the resulting trajectory had to be stabilized with a state-of-the-art algorithm to be successful. In addition to testing motion generated from high specifications to the stabilized motion in simulation, we express visual features at Whole B

    Source
  • TALOS models for Stack of Tasks

    Research Publication - The Stack of Tasks TALOS collection supplies plain URDFs, launch scripts and flexible-joint simulation support for motion generation and planning. It includes a Gazebo spring plugin and configurable starting poses.

    Source

Resources

Media library

Official exact-product technical material, open-source software and first-party media for TALOS. Generic humanoid coverage is excluded.

Timeline

  • PAL Robotics integrates Orientus sensors in TALOS and Kangaroo

    2024-03-26 - Capability - Orientus supplied inertial and attitude measurements at 1 kHz with low latency. PAL combined base-pose measurements with joint kinematics and ground-contact information for whole-body torque control. The compact sensor was integrated in both TALOS and Kangaroo; the platforms were used for different locomotion research.

    Source
  • PAL Robotics demonstrates TALOS climbing uncharted stairs

    2023-12-30 - Demonstration - A lidar-derived elevation map fed a contact planner and model predictive controller, with inverse dynamics computing joint torques. TALOS climbed 15 cm stairs without prior terrain information and adapted after the stairs were moved. The completed ROS architecture ran onboard TALOS; applying the walking work to Kangaroo remained future work.

    Source
  • PAL Robotics TALOS joins University of Waterloo RoboHub

    2018-11-07 - Deployment - The University of Waterloo described TALOS as the first platform of its kind in Canada. RoboHub planned to use TALOS for control and gait synthesis, manipulation planning, HRI, machine learning and high-DoF robot research.

    Source

Context

Market context

TALOS is PAL Robotics' full-size humanoid research robot for bipedal locomotion, whole-body control and manipulation.

  • Target environment: Robotics labs, humanoid research centres and controlled indoor testing environments.
  • Workflow context: Bipedal locomotion, whole-body control, manipulation, balance, perception and humanoid software development.
  • Deployment model: Research humanoid platform used by academic and industrial robotics teams.
  • Commercial maturity: Established research humanoid platform from PAL Robotics.
  • Adoption constraints: Requires robotics expertise, controlled lab conditions, safety procedures and advanced software development.
  • Market position: High-end humanoid research robot for teams working on locomotion, control and manipulation.
  • Adjacent products: Full-size humanoid platform for advanced research in walking, manipulation and whole-body control.

Comparable Systems

  • TORO

    DLR Institute of Robotics and Mechatronics - Anthropomorphic Biped

  • HUBO2

    Rainbow Robotics - humanoid robotics

  • E1

    Noetix Robotics (松延动力) - humanoid robotics

  • PM01

    ENGINEAI - humanoid robotics

  • MagicBot Z1

    MagicLab - Anthropomorphic Biped

  • Asimov 1

    Menlo Research - Humanoid Robot

Related Systems

  • T800

    ENGINEAI - humanoid robotics