Open source robots: to make, or not to make?

Abstract¶

This short presentation recalls the principles behind The Open Source Definition and discusses how we should adapt them to the more general context of open source robots. It is also an invitation to check out and update the Awesome Open Source Robots list, where we list robots and check the features that make them open source: license? Build instructions? Open source software?

Content¶

	Slides
	Awesome Open Source Robots
	Pinocchio tutorial

Pinocchio tutorial¶

This tutorial gave a preview of Pinocchio 3 with a focus on combining Pinocchio with CasADi to solve robotics problems such as inverse geometry and trajectory optimization. For example, writing a CasADi computation graph to solve inverse geometry with Pinocchio looks like this:

import casadi
from pinocchio import casadi as cpin

cmodel = cpin.Model(model)
cdata = cmodel.createData()
q_sym = casadi.SX.sym("q", model.nq, 1)
cpin.framesForwardKinematics(cmodel, cdata, q_sym)

error_function = casadi.Function(
    "end_effector_error",
    [q_sym],
    [
        cpin.log6(
            cdata.oMf[end_effector_id].inverse() *
            cpin.SE3(transform_target_to_world)
        ).vector
    ],
)

opti = casadi.Opti()
q_var = opti.variable(model.nq)
cost_function = casadi.sumsqr(error_function(q_var))
opti.minimize(cost_function)
opti.solver("ipopt")
sol = opti.solve_limited()
q_sol = opti.value(q_var)

The tutorial ships a binary pre-release of Pinocchio 3 that can be installed by following these instructions.

Jupyter notebooks¶

	Installing Pinocchio 3's preview
	Inverse geometry
	Trajectory optimization
	Constrained dynamics
	Working with obstacles

Slides¶

	Overview of Pinocchio
	Installing Pinocchio 3's preview
	Inverse geometry
	Trajectory optimization
	Constrained dynamics
	Collision detection

About JNRH¶

JNRH are one of the main French-speaking forum for the academic and industrial community studying humanoid and legged robots, biomechanics, and complex polyarticular systems that interact physically and cognitively with their environment. The goal of this forum is to bring together the national research community to discuss the various scientific topics related to these fields: mechanical and mechatronic design, modeling and simulation, control, numerical optimization, planning, learning, perception, motion analysis, etc. Doctoral students, postdoctoral researchers, and scientists are invited to present their latest work and research interests in the context of open scientific discussions. JNRH are also an opportunity to exchange ideas about the organization of our community, particularly in preparation for the IEEE/RAS Humanoids conference that will take place in Nancy in 2024.