Services | PickNik

MOTION PLANNING

IDENTIFICATION OF BEST PATH PLANNING APPROACH

□

Consulting on best algorithm to use for a given application and problem type

□

Benchmarking of competing approaches for analytical analysis

□

Implementation of novel algorithms and approaches

□

Literature reviews and academic surveys

□

Technology readiness level analysis and reporting

PROBABILISTIC/SAMPLING BASED METHODS

□

Consulting and development of the Open Motion Planning Library (OMPL)

□

Multi-query planners including PRM, LazyPRM, PRM*

□

Single-query planners including RRT, RRTConnect, RRT*, T-RRT, and LazyRRT

□

Asymptotically near-optimal planners including SPARS, SPARS2

□

Projection-based methods including EST, KPIECE, AtlasRRT

OTHER MOTION PLANNING METHODS

□

Search Based Planning Library (SBPL)

□

Stochastic Trajectory Optimization for Motion Planning (STOMP)

□

Covariant Hamiltonian Optimization for Motion Planning (CHOMP)

□

Dynamic Motion Primitives

EXPERIENCE-BASED MOTION PLANNING

□

Recall and repair of past plans for faster future planning

□

More deterministic results

□

Experts in Thunder, Bolt, and Lightning algorithms

ADVANCED INVERSE KINEMATICS

ANALYTIC SOLUTIONS

□

Experts in both computationally and manually generated solutions

□

Closed-form analytic solutions return complete solutions in microseconds

□

6 degrees of freedom (DOF) and less solvers

□

Consultation on mechanical design of robots

ITERATIVE METHODS

□

Advice in choosing the best inverse kinematics solvers for your application

□

Experts with off-the-shelf open-source KDL, LMA and Trac_IK solvers

□

Customize inverse Jacobian methods, gradient projection method, heuristic method etc.

□

Integrate complex constraints and behaviors in the null space

MULTIPLE ARMS

□

Implementing IK solvers for arbitrary robot configurations with multiple arms and legs

□

Support for with overlapping (shared) joints such as torsos

□

Inverse kinematic solvers that satisfy multiple constraint functions such as Stack of Tasks and particle swarm optimization

□

Motion planning with underconstrained waypoints that iteratively run IK

REAL-TIME CONTROL

CONTROL FRAMEWORKS

□

Integration and customization of the ROS Control Framework

□

Creation of custom controllers including position, velocity, and force controllers

□

Integration of the OROCOS control framework

REACTIVE PLANNING

□

Fast planners that are able to react and adjust to a changing environment

□

Dual mode Cartesian and free-space planning algorithms for wide range of applications

□

System wide performance benchmarking

COLLISION AVOIDANCE

PLANNING AROUND STATIC AND DYNAMIC OBSTACLES

□

Modeling of static and dynamic environments as “Planning Scenes”

□

Collision aware planning using virtual maps of the environment

□

Avoidance of local minimum around obstacles while planning

□

Experts with relevant libraries including the Flexible Collision Library (FCL), Octomaps, and the Point Cloud Library (PCL)

ROS FRAMEWORK DESIGN

SOFTWARE ARCHITECTING WITH ROS

□

Consulting on best practices for standard ROS paradigms

□

Training for internal development teams new to ROS

□

Historical insights into design decisions and ongoing ROS shortcomings

ROS2 DDS MIDDLEWARE CONVERSION

□

Assistance with converting legacy ROS packages to ROS2

□

Expertise in ROS2 advantages and release status

HIGH-QUALITY ROS PACKAGES / STACKS

□

Fully ROS compliant code repository setup and catkinization

□

Documented, tested, and code-reviewed deliverables

□

Continuous integration scripts and integration testing

□

Nvidia GPU-enabled Docker containers for multi-platform support

□

ROS C++ style guidelines, roslint, and catkin lint enforcement

2D NAVIGATION

MOBILE BASE NAVIGATION STACK

□

ROS navigation stack integration, customization, and deployment

□

Mobile base path and trajectory planning

□

OpenSlam's Gmapping, Google Cartographer algorithms for SLAM-based tracking

□

ROS-Control framework for switching between simulated a real-world environments

□

Simulated Gazebo system of robot base navigation

VIRTUAL REALITY

IMMERSIVE ROBOT VISUALIZATIONS

□

Iterate faster on hardware by testing new designs virtually before physical prototyping

□

Deploy high situational awareness teleop environments

□

Use ROS Rviz with off-the shelf 3D headsets like the HTC Vive and Occulus Rift

□

Evaluate and plan your collaborative robot integration strategy in virtual reality

□

Run Unity3D on Windows or Linux

ROBOT MODELLING

ASSEMBLY MODELLING

□

Setup Solidworks assembly files to have proper mechanical mates and joint limits

□

Setup coordinate systems for DH parameterization

□

Create Interface Control Documents for consistent representation of the robot kinematics

URDF GENERATION

□

Convert CAD files to compliant URDF specification

□

Verification of joint limits and appearance in Rviz

□

Integrate inertial properties for simulation in Gazebo

□

Integrate actuator and controller properties for ROS Control

CAD PACKAGES

We are currently compatible with the following CAD packages:

□

Solidworks 2017 / 2018

□

OnShape

MACHINE LEARNING

IDENTIFICATION OF BEST MACHINE LEARNING APPROACH

□

Consulting on best machine learning approaches to use for various applications

□

Experience with wide variety of machine learning algorithms for a wide variety of tasks

□

Implementation of novel algorithms and approaches

□

Technical road-mapping

PERCEPTION PIPELINES

□

Neural Network based perception systems

□

SVD, Random Forests, Naive Bayes, etc...

□

Sophisticated algorithm analysis

□

Structured and Unstructured learning

DATA PIPELINE MANAGEMENT

□

Dataset Generation and Labeling

□

Feature Extraction and Data Munging

□

ETL Pipelines using AWS and other Cloud based services