Progress for week 43 (2019)

From Robin

Revision as of 10:39, 18 October 2019 (view source) Hpnguyen (Talk | contribs) (→Budget) ← Older edit		Revision as of 11:55, 18 October 2019 (view source) Emmaste (Talk | contribs) Newer edit →
Line 12:		Line 12:
	=== Accounting ===		=== Accounting ===
	** NO ACCOUNTING AVAILABLE **		** NO ACCOUNTING AVAILABLE **
		+
		+	== Emma ==
		+	=== Budget ===
		+	* Apply poet to the modified bipedal walker environment
		+	* Run poet with learning leg length on a powerful computer (Need computation power?)

---

Revision as of 11:55, 18 October 2019

Contents 1 Tony 1.1 Budget 1.2 Accounting 2 Emma 2.1 Budget

Tony

Budget

-Plotte posisjon, Joint angles ,fart i matLab for å se om bevegelsen som predikeres er noe forholdsvis lik MoCap-data. 
-Prøve en approach hvor man velger minisett av kroppsdeler på MoCap data til kreative robot topologier.               
-Finne et godt mål for kvantitativ analyse av bevegelser
-Kinetisk energi, akkumulerte distansen for hvert ledd...
-Dele opp MoCap-data inn i subsett og observere om man får bedre resultater mellom MoCap og det predikerte. (x5 nevrale nett, bare Dense) 

• Variational autoencoder for å sample fra latentspace til unike robot topologier.

-Roboten beveger seg da noenlunde, men her kan det tenkes å utnytte en enkel evolusjon for å maksimere fitness/bevegelsen/energien

Accounting

• • NO ACCOUNTING AVAILABLE **

Emma

Budget

• Apply poet to the modified bipedal walker environment
• Run poet with learning leg length on a powerful computer (Need computation power?)